Comments Received during the Public Review Period on the
Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2012
Commenter: Giles Ragsdale
AECOM
Comment: Regarding chapter ES.2. - My opinion is that under this recent trends paragraph, note
should be taken of the results of Figure ES-15. There is a positive story to tell in that despite
increased population and Real GDP, emissions per capita and per $GDP have been trending
downward since 1990 and by more than a negligible amount.
Comment: Regarding figure ES-3: I think the title should be revised. I might be confused, but I
do not think the data on the graph reflect the title of the figure. I see that each year's annual total
compared to 1990 is represented, but I do not see the "Cumulative Change" noted in the title.
For cumulative change, it would seem that 1991 would be -25 as noted, but 1992 would be 52,
1993 would be 261, etc.
Commenter: William Herz
National Lime Association
Comment: In response to the last iteration of EPA's Greenhouse Gas Inventory Draft, published
in March of 2013, NLA submitted comments that recommended EPA discontinue using the
IPCC emission factors to account for LKD emissions, and that the agency also take into account
C02 emissions from off-spec lime, scrubber sludge, and other wastes. A copy of NLA previous
comments is included in Attachment 1. This issue continues to be important to NLA members,
not only to help ensure the completeness and accuracy of the data EPA publishes but also to
ensuring the achievement of EPA's stated goal of agreement and alignment with the GHG
mandatory reporting system.
Currently, EPA calcination emission calculations rely solely on output-based emission factors
from the IPCC 2006 GHG Guidelines, which we believe are outdated. Central to the NLA's
previous comments were recommendations to adopt accurate calcination emissions calculation
methodology for:
• Lime Products; and
• Lime Kiln Dust (LKD); and
• Off-spec lime, scrubber sludge and other wastes.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
1
-------
Each of our recommendations was based on analysis of actual production data, including CaO
and MgO oxide contents of lime and LKD, which had been provided to NLA by member
companies. These results of this data were incorporated in the comments NLA submitted to your
office last year. The comments, together with the data we provided, should be sufficient to
provide EPA with the basis to generate more accurate emissions estimates for LKD and other
lime products (including off-spec lime and scrubber sludge).
In sum, NLA's earlier comments concluded that while the IPCC's output-based approach for
estimating calcination emissions from U.S. lime products may be accurate as to the overall data
to be published, it nonetheless understates emissions from LKD and other byproducts and wastes
generated in the United States. For that reason, NLA recommended that lime calcination
emissions be multiplied by a factor of 1.06 (not 1.02) to account for LKD and a factor of 1.02 to
account for wastes generated at lime plants; neither of these are currently accounted for which
we believe is a critical error.
When the current Draft Inventory of U.S. Greenhouse Gases and Sinks 1990-2012 was published
on February 21, 2014, it was disappointing that none of our recommendations concerning the use
of more accurate correction factors had been adopted and EPA continued to rely on the outdated
IPCC factor of 1.02 to account for LKD. Likewise, EPA took no action in relation to off-spec
lime and other wastes.
As we stressed in our previous comments concerning the earlier draft, NLA conclusions and
recommendations were premised on our belief in the need for EPA's published data to be
accurate; especially when NLA's members are willing to supplement the agency's data with
accurate data of their own. Because EPA relies solely on the questionable IPCC LKD generation
rates, calcination emissions continue to be understated. Accordingly, we again urge EPA to adopt
our recommendations; if there are other supporting data we can also provide that would add
further weight to and/or support for our recommendations, please let us know.
In addition, we recognize that EPA has a substantive interest in having both the GHG Inventory
and the Mandatory GHG Reporting system be in agreement as much as possible. This is
important not only for EPA's credibility but also for the public's and stakeholders'
understanding of these issues. In this regard, as we stated in our previous comments:
Lime Kiln Dust
"...based on data reported to NLA from our members, emissions from generating LKD account
for about 6% of calcination-related emissions from lime manufacturing (in 2011, it was 5.8%).
Currently the IPCC multiplies lime product-related emissions by a "correction factor" of 1.02 to
account for LKD. The IPCC Guidelines acknowledge that this correction factor for LKD is
borrowed from its chapter on cement, which in turn explains that the factor for cement kiln dust
(CKD) is relatively low because most CKD is recycled back into the process.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
2
-------
By contrast, the lime industry does not recycle LKD back into the process, and thus borrowing
such a factor to account for LKD-related calcination emissions is inappropriate.
EPA's reliance on the IPCC's LKD generation rate of 2% (rather than 6%) understates
calcination emissions from our members alone by 535,610 tons. This is roughly 5.4% of our
members' total emissions, and twenty times the understated calcination emissions described
earlier for lime products."
Off-Spec Lime. Scrubber Sludge, and Other Wastes
"The IPCC Guidelines do not to take into account calcination emissions resulting from wastes
commonly generated at lime plants (e.g. off-spec lime that is not recycled, scrubber sludge).
Again, based on 2011 data reported to NLA from our members, calcination emissions from
production of such wastes account for approximately 1.7% of total calcination emissions, or
256,000 tons. To address this omission, we recommend that EPA multiply quicklime calcination
emissions by a factor of 1.02."
Conclusion:
NLA believes the deficiencies in the proposed inventory are significant and should be corrected.
In the aggregate, EPA has underestimated lime emissions by approximately 814,000 C02 tons;
as the off-spec materials generate 256,000 tons (completely unaccounted for in the inventory)
and 535K tons (the difference in LKD emissions when utilizing the correct emissions factor;
(854K - 319K)). This represents an underestimate of approximately 5.1%, which is not
insignificant.
Commenter: Marlen Eve
USDA Animal and Plant Health Inspection Service
Comment: Executive Summary:
Page 1 lines 29-30: Excellent!
Page 2 lines 9-10: Needed for effective comparison.
Page 5 Figure ES-3: Very impressive and encouraging trend!
Page 10 lines 10-14: This is excellent - it enables an accurate sectorial picture otherwise difficult
to estimate.
Page 14 lines 27-31: Noteworthy point that technology improvements can be so effective in this
area.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
-------
Page 15 lines 9-10: Good to see this point made in Summary - an area USDA is researching and
still in need of improvement.
Page 16 lines 19-21: Suggests an area in need of more oversight and regulation in a fast growing
industry.
Page 17 line 13: Noteworthy effect of improved land-use and forests as sink. Question - why
have these sinks not increased since 2007 - compared to notable improving trend over 1998-
2004?
Page 20 lines 7-10: Good point to note - not sure this is widely recognized - and how difficult it
is to manage for lower levels. This is a clearly-needed USDA research area.
Page 20 lines 29-30: Good to mention in Executive Summary given this is a problematic area in
many developing counties including China and India.
Page 27 lines 7-9: Good point to make - it identified an area that could benefit from future
research.
Page 5 line 27: Is this very long table needed in Executive Summary?
Page 11 line 15: Reference that low fuel prices during period 1990-2012 in part contributed to
increase in number of vehicle miles. Hopefully this can be substantiated through economic
comparison - I think fuel prices increased considerably during this period relative to other
consumer prices. And when I look at some internet sites such as:
http://www.inflationdata.com/inflation/images/charts/Oil/Gasoline_inflation_chart.htm, they
seem to reflect that the statement that gas cost has remained low and thus the conclusion that this
leads to increase in number of vehicle miles could be challenged.
Page 17 lines 17-18: Is this very long table needed in Executive Summary?
Comment: Introduction:
Page 13 line 1: Very good summary of all input sources of data and expertise in one diagram!
Comment: Agriculture:
Page 1 lines 5-8: Very good way to focus on what is critical in agriculture practices!!
Page 1 lines 16-20: We liked this up-front summary and focus on what is critical!
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
4
-------
All pages: There is a great abundance of numbers, informational statements, tables and some
figures. The details can be overwhelming but we view the text and supporting data as essential,
comprehensive, well-balanced, and superbly organized by easy to read, consistent sections on
each source of non-C02 GHG. The methods used should be of value to other countries as a 'role
model' on what data is needed and on how to assess uncertainty and apply verifications and
recalculations.
Overall—comprehensive and well-written chapter on a difficult subject.
Several locations in the Chapter: There are references to number of cattle/dairy cattle
increasing/decreasing but overall CH4 emission increasing due to digestibility. I can look at
some trends in the NASS that would indicate that the trends in numbers of cattle stated don't
quite coincide with my quick review of NASS. But they do state a lot of adjustments that they
made to the numbers that I don't have the time to work through. And I definitely don't have the
background on digestibility - few, if any, in Veterinary Services would. So I can't validate or
refute, and would not want our brief review to be considered a "peer review". I would hope that
this section and others in the paper have been appropriately peer reviewed to avoid any improper
conclusions developed which could have an undue negative influence on animal agriculture.
Comment: Land Use, Land-Use Change, and Forestry:
All pages: As commented on Agriculture Chapter we note that there is a great abundance of
numbers, informational statements, tables and some figures. The details can be overwhelming
but we view the text and supporting data as essential, comprehensive, well-balanced, and
superbly organized by easy to read, consistent sections on each source of GHG. The methods
used should be of value to other countries as a 'role model' on what data is needed and on how to
assess uncertainty and apply verifications and recalculations.
Overall: Comprehensive and well-written chapter on a difficult subject. This category is
especially important to developing countries where land use is in flux and where practices such
as forest cutting and clearing, fire use, and extensive degradation by grazing is wide-spread.
Comment: Recalculations:
Page 1 lines 2-4: We felt this is one of the most important chapters in the Report given it
provides a protocol and verification annually of the estimates. It has the salutary benefit of
credibility of estimated made given they are constantly under re-evaluation as new data (past and
present) and methods are developed and accessed. Some of the changes appear large in
magnitude - but this may not be unusual where only imprecise data was available initially.
Possibly add a summary or a tabulation of what this report achieved in the way of new data, new
methods and new findings that were not mainstream in prior analyses and thinking.
We note with interest some prior assumptions (or simple lack of information or awareness) on
aspects of agriculture and land use / forestry of special interest. Some of these new perceptions
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
5
-------
are now and in the immediate future will be important in on-going and future negotiations over
land and land practice monitoring by different countries.
Commenter: Marlen Eve
USDA Agriculture Research Service
Comment: Agriculture chapter: Pagel line 6:
Seems to imply C removal is only related to land-use change. Assuming other management does
not have an impact?
Comment: Agriculture chapter: Page 1 line 18:
"other cropping practices" a little vague - such as?
Comment: Agriculture chapter: Page 2 lines 16-17:
"These non-ruminants emit significantly less CH4 on a per-animal-mass basis than ruminants
because the capacity of the large intestine to produce CH4 is lower."
Add ... lower than in a rumen.
Comment: Agriculture chapter: Page 9 line 20:
"... increasing use of liquid manure management systems, which have higher potential CH4
emissions than dry systems." Are there any estimates on the adoption of methane capture from
liquid manure?
Comment: Agriculture chapter: Page 29 line 3-4:
This sentence needs to be rewritten without all the "nots:"
"However, renewal of pasture that is not rotated with annual crops occasionally is not common
in the United States, and is not estimated."
Comment: Agriculture chapter: Page 30:
In general DAYCENT appears to perform well, but recent work by Campbell et al., 2014
suggestion DAYCENT may underestimate N20 emissions. "Overall, DAYCENT performed
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
6
-------
well at simulating stover yields and low N20 emission rates, reasonably well when simulating
the effects of management practices on average grain yields and SOC change, and poorly when
estimating high N20 emissions. These biases should be considered when DAYCENT is used as
a decision support tool for recommending sustainable corn stover removal practices to advance
bioenergy industry based on corn stover feedstock material." (Campbell et al., 2014). Thus, as
more empirical data becomes available it could be used to improve DAYCENT.
Comment: Agriculture chapter: Page 41 Table 6-27:
Key Assumptions for Estimating Emissions from Field Burning of Agricultural Residues: Is it
correct that this is refers only to residue that are burned in the field or does it include residues
harvested and burned for energy - clarify.
Commenter: Other
USDA
Comment: Page 6-7, line 18: "months" should be inserted after "4-6."
Commenter: Carrie Reese
Pioneer
Comment: Gas Well Com pi eti on s/Workovers with Hydraulic Fracturing:
Pioneer commends EPA's consideration of stakeholder comments to the 2013 Inventory and
subsequent development of control technology-specific, net emission factors for gas well
completions and workovers with hydraulic fracturing. This approach makes use of a more
comprehensive data set and provides greater transparency regarding EPA's accounting of
emissions reductions carried out by the industry. However, Pioneer feels that this methodology
can still be improved upon.
Emissions quantified in the Greenhouse Gas Reporting Program (GHGRP) for 2011 and 2012
are based on engineering estimates and best available monitoring methods (BAMM) in addition
to direct measurements. In Pioneer's initial review of 2011 and 2012 GHGRP data for "HF
completions that vent", average emissions per event (Mg CH4) computed by an estimation
methodology appear to be nearly tenfold that of directly-measured emissions. Until there is
further understanding of the nature of these events, Pioneer suggests that EPA develop control
technology-specific, net emissions factors focusing on measured data from the GHGRP and
measured data contributed by other accepted sources.
Published by the University of Texas at Austin in September 2013, Measurements of Methane
Emissions at Natural Gas Production Sites 1 (Allen, et. al) quantifies emissions from 27 gas well
completions in multiple production regions. Representative gas well completions from nine
operators, which conduct about half of all new well completions, were sampled. The
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
7
-------
measurement results, a product of peer-reviewed, scientifically-robust, and fully-disclosed
methodology, present a basis to consider the reasonableness of other data provided under less
controlled circumstances.
Referred to in the 2014 Inventory, the November 8, 2012 document entitled "Inventory of U.S.
Greenhouse Gas Emissions and Sinks: Updating Emissions Estimate for Hydraulically Fractured
Gas Well Completions and Workovers", identifies four categories of events and proposes new
emission factors for each of these categories. The following table provides a comparison of the
EPA emission factors in the 2014 Inventory to the measurements reported by Allen, et al.
(2013). The study reports emissions data for completion flowbacks only, with no measurements
for workovers with hydraulic fracturing.
Type of completion flow back EPA
or workover Emission
Factor
Observed Emission
Factor from Allen, et
al, (2013)
Wells that vent without flaring or
Reduced Emission Equipment (REC)
41
0.83 (observed mean)
OS-124 (range)*
Wells that flare (without REC)
5
Not observed
Wells with REC that do not flare
3
4
Wells with REC that flare
6
15-18
*The wells that vented without flaring or REC observed by Allen, et al. (2013) had much lower potential
emissions (083 Mg) than the average potential emissions for all of the observed wells (124 Mg). Ifthe wells in this
category observed by Allen,et al.are representative of national populations of this category of wells, then the
emission factor forthis category would be0.8 Mgperevent; incontrast, ifthe sample of all wells observed by Allen,
et al. (2013) is considered representative of this category and it is assumed that the deployment of REC equipment
is random among all of these wells, then the emi ssion factor for thi s cate gory would be 124 Mg per event.
Comment: Liquid Unloadings:
In previous comments to the 2013 Inventory, Pioneer expressed support for EPA's development
of net emissions factors for liquid unloading events, but also noted concern that Subpart W
calculation methodology may tend to overstate emissions. Pioneer requests that EPA continue to
consider improvements to the calculations in this emissions category.
In the study referenced above, Allen, et al. (2013) also reported on emissions from liquid
unloadings.
The sample set of nine manual unloadings proved insufficient to allow for extrapolation at a
national scale, and the study team is conducting additional measurements to supplement the data
collected in the first part of the study. However, Allen, et al. (2013) does report an important
observation from the initial effort, demonstrating that the Subpart W methodology for liquid
unloadings without plunger lifts (based on engineering calculations and not direct measurements)
overestimates emissions for every measured event. Collectively, emissions are estimated five
times higher than the measured emissions.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
8
-------
Pioneer again suggests that underlying causes for overestimation of emissions may be EPA's
assumption that a full wellbore volume of gas is vented with each unloading and the assumption
that a well unloads for one hour on average. In addition, Allen, et al. (2013) observed
intermittent flow rates during unloading events and proposes that EPA's assumption of a
continuous gas flow rate may be another contributor to overestimation .
Comment: Pioneer applauds EPA's commitment to refine emission estimates in the U.S.
Greenhouse Gas Inventory to reflect the best available information. As the body of scientific and
engineering knowledge continues to grow with regards to emissions from oil and gas activities,
Pioneer contends that emphasis must be placed on directly-measured data and that results from
these direct measurements should in turn inform corresponding estimation methodologies .
Collaboration is the key to this process, and Pioneer looks forward to continued dialogue with
EPA.
Commenter: Chris Busch
Energy Innovation: Policy and Technology
Comment: The EPA should take steps to address clear evidence that its inventory of GHG
emissions is undercounting methane. In the short run, as part of finalizing the 2014 inventory, the
agency should make the case for a significant effort to improve the inventory of emissions from
the natural gas sector. In the longer run, the agency should develop a plan for integrating top-
down data as well as new technologies that operate at ground level that can assist in leak
detection and measurement. The federal government should be placing more emphasis and
devoting more resources to this effort.
Comment: Brandt et al.'s work illustrates the value of top-down measurements to provide
evidence of overall emission levels over large areas. The EPA should move to collect airborne
measurements into its GHG inventories. By conducting measurement campaigns, EPA will be
able to obtain atmospheric data that is more comprehensive across space and time. This will
enable the agency to identify aggregate emissions levels with much greater accuracy and will
help to improve confidence intervals. Current confidence intervals are much too small.
Comment: Emerging technologies can link emissions back to sources, enabling the EPA to
conduct an effective ground-level measurement campaign. Infrared cameras are effective at
locating leaks, and their use has been required under a recently approved Colorado regulation.
Low cost stationary detectors are also under development. The newest detectors can locate leaks
and estimate their magnitude from a distance, which reduces the challenge of acquiring property
owner permission that bedevils direct on-site measurement.
Comment: The current oil and gas boom has been unleashed by a wave of technological
innovation (directional drilling, hydraulic fracturing, and other emerging techniques, like
"acidizing"). Governments need to keep pace with faster innovation on the regulatory side. New
monitoring technologies are an opportunity for greater accuracy, and the EPA should move
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
9
-------
quickly to use these technologies to transform government monitoring of emissions. Better
monitoring of emissions will help the EPA solve the mystery of the missing methane and provide
the best objective guidance to policymakers, regulators, and society.
Commenter: Kerry Kelly
Waste Management
Comment: WM is pleased to see that the emissions from the waste sector and landfills in
particular, continue to trend downward, while methane emissions control via gas collection and
combustion in renewable energy projects or flares continues to grow. We also noted with
interest, the discussion of planned improvements to measuring landfill emissions by replacing
the default 10 percent oxidation with a more accurate, science-based estimate. It is this aspect of
the Draft Inventory that is the subject of our comments.
The Agency refers to a growing body of peer-reviewed literature describing both field and
laboratory studies that all indicate that the default oxidation value of ten percent understates the
oxidation rates achieved at well-managed landfills. EPA's careful analysis of peer-reviewed
literature and field measurement studies resulted in recent regulatory changes to estimating
methane oxidation in landfill cover in the Greenhouse Gas Mandatory Reporting Rule (GHG
MRR). These changes allow greater use of site-specific conditions rather than national default
assumptions and will greatly increase the accuracy of landfill facility methane emissions
estimation. We urge that the Agency also update its national inventory methods to reflect these
changes, so that it can improve the accuracy and reliability of the U.S. GHG Inventory.
Comment: The EPA's Decision to Revise the Methane Oxidation Factor Used in the GHG MRR
is Well Supported by Peer-Reviewed Science:
Numerous studies have been conducted worldwide and referenced in the scientific literature that
address and document methane oxidation in cover soils, as well as gas collection efficiency. In
2009, The Journal of Environmental Quality published a comprehensive literature review. The
paper references over 60 technical documents dating from 1960 to the present, with the majority
of the papers being published in the 1990s and 2000s. Overall, based on review of 42
determinations of the fraction of methane oxidized in a variety of soil types and landfill covers,
the mean fraction of methane oxidized across all studies was 36 percent with a standard error of
6 percent. For a subset of 15 studies conducted over an annual cycle, the fraction of methane
oxidized ranged from 11 percent to 89 percent with a mean value of 35 percent + 6 percent,
nearly identical to the overall mean.
In July 2007, the Solid Waste Industry for Climate Solutions (SWICS) released its first white
paper titled Current MSW Industry Position and State-of-the-Practice on LFG Collection
Efficiency, Methane Oxidation, and Carbon Sequestration in Landfills (White Paper). The
public and private members of SWICS shared the White Paper with EPA as it developed the
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
10
-------
GHG MRR requirements for evaluating and reporting MSW landfill emissions. In January 2009,
SWICS updated the White Paper to incorporate additional studies noted above.
Since the release of the 2009 White Paper, a number of studies have been published in peer-
reviewed literature, most notably an entire special issue of the journal Waste Management (2011)
on Landfill Gas Emission and Mitigation sponsored by Consortium for Landfill Emissions
Abatement Research (CLEAR). CLEAR is an International Waste Working Group (IWWG)
Task Group, which focuses on landfill gas emission to the atmosphere. The group has members
from 12 countries, across four continents. A number of papers in the special issue focus on the
use of compost biocovers, bio-windows or permeable gas dispersion layers to treat and oxidize
landfill gas in situ (Huber-Humer et al., 2011; Pedersen et al., 2011; Scheutz et al., 2011;
Pawlowska et al., 2011; Dever et al., 2011; and Jung et al., 2011). Additionally, several papers
in the special issue, Ranchor et al., (2011); Abichou et al., (2011) and Chanton et al., (201 lb),
examined the response of the methane oxidizing microbial community to methane loading to the
cover soil. Two key papers, Bogner et al., (2011), and Spokas et al., (2011), describe recent
work in California where field measurements of emission and oxidation were coupled with
extensive modeling efforts. Chanton et al., (201 la) published the results of 37 seasonal sampling
events at 20 landfills with intermediate covers over a four-year period. Abichou et al. (201 lb)
examined the best approach towards describing central tendencies in oxidation data and reported
that the results were generally distributed normally so that mean values could be used.
In November 2012 SWICS, with the participation of Dr. Jeffrey Chanton of Florida State
University and Dr. Morton Barlaz of North Carolina State University, finalized an addendum
(2012 Addendum) to the Methane Oxidation section of the 2009 White Paper. The 2012
Addendum includes methane oxidation results from evaluations of 90 landfills as compared to
the 47 published evaluations available in 2009.
In reviewing and incorporating the results of these peer-reviewed studies of landfill methane
oxidation, the 2012 Addendum updated the 2009 White Paper results as follows:
1. Clay cover: The number of studies in clay cover increased from five in 2009 to 31 in
2012. The mean fraction of methane oxidized increased from 18 percent to 30 percent, while the
median fraction oxidized increased from 14 to 29 percent.
2. Sandy soils cover: The number of studies in sandy soils doubled from eight to 16, with
the mean oxidation value changing very little (55 to 54 percent) while the median value
increased from 43 to 50 percent methane oxidized.
3. "Other" covers: The number of studies in "other" cover soils increased by nine and both
the mean and the median fraction oxidized values increased slightly.
4. The overall mean oxidation value across all of the studies increased from 35 to 38 percent
while the overall median oxidation fraction increased from 31 to 33 percent.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
11
-------
Comment: The SWICS 2012 Addendum Definitively Supports a Significant Increase to the
Current Default Value of Ten Percent:
The 2012 Addendum concluded that the EPA default oxidation value of 10 percent
underestimates typical methane oxidation and is not representative of expected methane
oxidation at sites utilizing organic, clay, sand or other cover types. EPA derived the default
value from one field study performed at one poorly maintained landfill with no gas collection
system, Czepiel et al. Analysis of the 90 studies highlighted in the 2012 Addendum indicates
that if a single value is considered for methane oxidation it should be between 33 and 38 percent.
The 2012 Addendum also examined methane oxidation as a function of methane loading to the
cover layer of the landfill. Recent studies show that the percent oxidation is an inverse function
of the rate of emission (Stern et al., 2007; Rachor et al., 2011; Chanton et al., 201 la,b). At lower
emission rates, the methanotrophs in the soil cover can consume a larger portion of the methane
delivered to them, potentially oxidizing 95 to 100 percent (Humer and Lechner, 1999, 2001a,
Huber-Humer 2008; Powelson et al., 2006, 2007; Kjeldsen et al., 1997). As flux rates increase,
the percent oxidation decreases and the methanotrophs can become overwhelmed with methane.
Thus, as methane emission increases, percent oxidation decreases (Powelson et al., 2006, 2007).
A mathematical model of cover oxidation developed by Dr. Tarek Abichou of Florida State
University (Abichou et al., 2010), demonstrates that at lower methane fluxes, oxidation rates are
equal to the methane loading to the soil cover. Oxidation keeps pace with flux, and the soil
cover is able to oxidize all of the methane coming from below. At lower loading rates, methane
oxidation is equal to 100 percent. As flux increases, the cover is not able to oxidize all of the
incoming methane, and the percent oxidation falls off. Therefore, percent oxidation starts to
decrease as the methane loading to the cover increases. This relationship is shown clearly in the
laboratory column studies of Rachor et al., (2011). Field studies have also confirmed this
relationship between methane flux and percent oxidation (Chanton et al., 201 la, b). At low
rates of methane emission, the percent oxidation is near 100 percent. As emission rates
increase, the percent oxidation decreases. This analysis served to support the approach that EPA
finalized for determining a more accurate methane oxidation fraction by calculating the methane
flux rate for the landfill.
In addition to the 2012 Addendum, the landfill sector provided data for 262 private and public
landfills reporting under Subpart HH. The dataset allowed the Agency to evaluate several
possible options for determining more accurate methane oxidation fractions. The data
conclusively showed that the average oxidation fractions for different soil cover types are all
well above the default 10 percent value required by Subpart HH, and underpin the need for a
revised default value or more refined method for determining an oxidation fraction at a site.
Comment: WM recommends that the Agency carefully consider its analysis underpinning its
decision to estimate facility-level methane oxidation by calculating the methane flux rate and
consider how that methodology could be used at the national inventory level. The work done by
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
12
-------
the Agency in updating methods for facility-level calculation of methane oxidation will greatly
improve the accuracy and reliability of emissions estimates for landfills. We urge EPA to
endeavor to make similar improvements to its national inventory of emissions and sinks at
landfills.
Commenters: Anna Moritz, Kevin Bundy, Sparsh Khandeshi,
Center for Biological Diversity and Environmental Integrity Project
Comment: We are concerned that the emissions factors and global warming potential for
methane that are currently employed by EPA have resulted in a significant underestimate of U.S.
anthropogenic methane emissions. First, multiple studies have reported far higher leakage factors
from oil and natural gas operations than EPA currently uses. It is essential that EPA arrive at
accurate numbers. In addition, even reported leakage numbers necessarily underestimate
emissions because they omit undiscovered and unreported leaks, such as those by smaller
operators. EPA should immediately commence on-the-ground data collection and, until the
results from these efforts are available, account for these factors by presenting a range of likely
underreporting. Second, the climate impact of methane is underestimated because the inventory
reports normalized methane emissions using solely a 100-year global warming potential
("GWP") and an outdated value for the GWP.
Methane is an important component of climate strategies to avoid Arctic disaster and other
catastrophic tipping points. Unlike other traditional greenhouse gases that have atmospheric
lifetimes of a century or more, methane remains in the atmosphere for only about 12 years. This
means that a reduction in emissions today will not only slow the increase in radiative forcing, but
also result in actual decreases in radiative forcing in a short time - just over a decade. When we
are considering how to address the collapse of the Arctic cyrosphere or avoid near-term tipping
points, methane and other short-lived climate pollutants present an opportunity for rapid
reductions in climate forcing.
Because methane mitigation is an important climate strategy, it is essential that the current
emissions levels from US sources be accurately characterized. This includes both emissions
factors for various industries and quantification using the most current values for global warming
potential: a 100-year GWP of 34 and a 20-year GWP of 86.
Comment: Emissions factors from oil and gas operations should be revised:
There is compelling evidence that leakage rates from oil and gas operations are far higher than
EPA emission factors suggest. For instance, Miller and colleagues recently used atmospheric
measurements to estimate that actual methane emissions are about 1.5 times larger than EPA
estimates, with fossil fuel methane emissions more than two times higher than estimated.
Observations from oil and gas operations in Colorado indicate that inventories underestimate
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
13
-------
methane emissions by at least a factor of two. Leakage rates over a Utah gas field were recently
estimated at 6.2 to 11.7%, well above the rates assumed by national inventories.
Moreover, EPA's data for oil and combined oil/gas wells omit the impact of hydraulic fracturing.
A recent white paper from Environmental Defense Fund summarizes findings from a number of
studies to conclude that emissions factors used in EPA's current inventory underestimate
methane emissions from oil wells that employ hydraulic fracturing.
Another major source of methane emissions from the oil and gas sector is leaks from pneumatic
devices. A recent study calculated emission factors for pneumatic devices to find that national
emissions from this source are likely at least twice the amount predicted using the emission
factors in the US GHG Inventory. This is another area where improvement of emission factors is
essential.
We urge EPA to consider the range of data available and update the emission factors that are
used in the GHG inventory to accurately reflect methane emissions from both venting and
leakage in the oil and gas industry. These data are critical as industry leaders and decision
makers consider mitigation options.
Recent reports have also substantiated an alarming rate of leaks from decaying gas pipeline
systems across the country, creating the need for systematic, on-the-ground data collection to
obtain an accurate quantification of emissions from this source. For example, according to a
recent study, the two distributors of natural gas in New York City and Westchester County
reported 9,906 leaks in their combined system for 2012 alone, and gas distributors nationwide
reported an average of 12 leaks per 100 miles of the 1.2 million miles of gas main pipes across
the country. More than 5,800 leaks were detected from aging gas pipelines underneath the streets
of Washington, D.C. These samples indicate that EPA's data are incomplete, and we urge EPA
to note this fact and undertake the efforts necessary to provide an accurate accounting next year.
Comment: The GHG Inventory should quantify methane emissions using AR5 GWPs:
EPA recently finalized technical amendments to the Greenhouse Gas Reporting Rule. These
changes included updating the methane GWP from the values in the IPCC Second Assessment
Report to those in the Fourth Assessment Report ("AR4") for reporting in year 2015 and beyond.
While this was an important improvement, we and other organizations joined Clean Air Task
Force in recommending that EPA utilize the most up-to-date science and adopt the most recent
methane GWPs from the IPCC Fifth Assessment Report ("AR5") as well as require reporting of
both 100-year and 20-year methane GWPs. 10 EPA declined to adopt the most recent estimates
of methane's GWP because current international reporting requirements under the United
Nations Framework Convention on Climate Change employ only 100-year GWPs and will
begin using AR4 GWPs in 2015.11
While we understand EPA's need to comply with international reporting requirements, we renew
our call upon EPA to consider updating the emissions reported in the U.S. GHG Inventory to
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
14
-------
reflect the AR5 GWPs, as well as report normalized emissions using both 20-year and 100-year
GWPs for methane. The US GHG Inventory is important domestically for both government and
private-sector decision-making and analysis. This is a purpose separate from international
commitments and requires more precise quantification of climate impacts. We appreciate the
inclusion of Appendix 6.1 in the draft GHG Inventory, which provides emissions estimates as
calculated with both AR4 and AR5 100-year GWPs. We ask that EPA make this information
more prominent so that users will be more likely to find and employ the updated emissions
estimates. Furthermore, it is important that EPA use the most accurate GWP for methane, which
includes carbon cycle feedbacks.
We further request that EPA consider reporting emissions using both the 100-year and 20-year
GWP for methane as this will allow the full consideration of climate consequences. The 100-year
GWP gives a better sense of how reductions can influence long-term climate stabilization, while
the 20-year GWP is useful when considering tipping points and near-term climate impacts.
Furthermore, the AR5 values for GWP have changed substantially since AR4. The AR5 methane
GWP of 34 is significantly higher than AR4 - 36 percent higher. The AR5 20- year GWP is 86
(19% higher than the AR4 GWP). These substantial increases in GWP mean that emissions data
reported using AR4 GWPs or earlier are understated. Accordingly, EPA must revise the GWPs
used in the inventory and ensure that they properly reflect carbon cycle feedbacks.
Comment: EPA Must Clarify Data Sources and Emissions from Biomass Electricity Generation:
According to the Draft Inventory, CO2 emissions from woody biomass and woody biomass
consumption (measured in trillion Btus) in the electricity generation sector increased nearly
tenfold between 2011 and 2012. It is not clear, however, how these emissions estimates were
derived. Although emissions of biogenic CO2 associated with electricity generation are reported
primarily for informational purposes pursuant to international accounting conventions, accurate
emissions data are critical to evaluating domestic renewable energy programs and accounting for
the actual climate consequences of increasing biomass energy generation.
The Draft Inventory states that biogenic CO2 emissions from the electricity generation sector data
were calculated using EPA's Clean Air Market Acid Rain Program dataset, while emissions from
other sectors were obtained from EIA's Monthly Energy Review. 13 An annex to the Draft
Inventory explains that "there were significant differences between wood biomass consumption
in the electric power sector between the EPA (2013) and EIA (2013) datasets." Accordingly, "the
electricity generation sector's woody biomass consumption was adjusted downward to match the
value obtained from the bottom-up analysis based on EPA's Acid Rain Program dataset."
The increase in emissions between 2011 and 2012, if accurate, represents a dramatic expansion
of emissions from this industry—nearly a full order of magnitude over the course of only one
year. It is impossible to discern, however, whether the Draft Inventory's emissions estimates are
either comprehensive or consistent.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
15
-------
The EIA Monthly Energy Review data used for other biomass emissions estimates does not show
a similar increase in woody biomass consumption between 2011 and 2012; in fact, these data
show a slight decline in both wood and other biomass "waste" consumption. The increase thus
must be reflected, if anywhere, in EPA's Clean Air Market dataset. Again, however, this is
impossible to discern because the full dataset does not appear to have been included or explained
further in either the Draft Inventory or the annexes.
Indeed, it appears that the AMPD dataset may be under inclusive of electrical generation
facilities using woody biomass as fuel. For example, a query performed on the AMPD website
for 2012 emissions data from all programs and all facilities returned 4,828 records nationwide,
only 23 of which list "wood" as the primary fuel source; CO2 emissions from these facilities in
2012, where emissions were reported at all, totaled only about 2.7 million metric tons. There are,
of course, more than 23 wood-burning power plants operating in the United States; indeed, there
are more than 23 such facilities in California alone, although no California plants appear in the
query report generated by the AMPD dataset. is Of course, if there are numerous biomass power
plants that are not listed in the AMPD dataset, use of this dataset for a "bottom up" emissions
estimate will likely underestimate emissions from this sector.
Given these apparent inconsistencies, EPA should clarify what data set it is using to estimate
biogenic CO2 emissions from electricity generation and should ensure that these data are
inclusive and comprehensive enough to produce an informative report.
Comment: Conclusion:
In sum, we commend EPA for compiling and reporting extensive data from various sources of
greenhouse gases within the United States. There remain, however, some areas where
improvements are needed to maximize the utility of the GHG Inventory for both international
reporting and informed domestic policy-making. First, emissions factors for the oil and gas
industries, including pipeline leakage, are very likely much too low to accurately reflect fugitive
methane emissions. Second, we request that EPA expand its reporting of methane emissions
using both 20-year and 100-year GWPs as well as report methane emissions in the main text of
the Inventory using the GWPs from AR5. And finally, we request that EPA clarify the sources
and accuracy of data used to estimate emissions from biomass combustion, particularly for the
electricity generation sector.
Commenter: Jeff Zimmerman
Damascus Citizens for Sustainability
Comment: Over the last several years it has become apparent that stray emissions of methane
from gas development projects across the United States are increasingly contributing to the
greenhouse gas levels and climate change. The purpose of our submissions today to your draft
inventory document is to bring to your attention a number of recent (2012-2013) studies and
reports providing actual measured emissions of stray methane from unconventional gas
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
16
-------
development using fracking. The data collected and reported in these publications documents a
range of additional scientific information that needs to be factored into the GHG emissions
inventory and the resultant impacts of climate change.
Three of these reports document extensive methane leakage from natural gas distribution
facilities in Manhattan, New York City, NY, (Payne and Ackley, March 2013) Boston, MA,
(Philips et al., 2012) and Los Angeles, CA (Peischl et al., 2013). The LA Basin report documents
methane leakage at 17% of total gas production in the LA Basin. Another report provides
methane leakage data in a natural gas production area in Wyalusing Township in Bradford
County, PA (Payne and Ackley, November 2013) and another report documents methane leakage
in fracked gas production areas of Leroy, Granville, and Franklin Townships in Bradford
County, PA (Payne and Ackley, 2012). A sixth report documents 4% methane leakage in the
Denver-Julesurg Basin in Colorado (Tollefson, 2012), and a seventh report documents a 9%
overall methane leakage rate from fracked gas development in the Uinta Basin in Utah
(Tollefson, 2013). These reports seriously call into question the much lower methane leakage
rates from fracked gas development estimated by EPA. A report by Miller and many others
summarizes the results of these and other similar studies and concludes that actual methane
leakage rates are almost five times the earlier EPA estimates (Miller et al., 2013). Each of the
reports we are providing with this comment letter should be included in the EPA inventory of
climate change and GHG data. The trend in these reports demonstrates that methane leakage
from unconventional gas development is far greater than previously thought. A comprehensive
reexamination of leakage rates and impacts is clearly required.
Commenter: Cynthia Finley
National Association of Clean Water Agencies (NACWA)
Comment: The wastewater treatment category includes publicly owned treatment works
(POTWs), septic systems, and industrial wastewater treatment systems. Although the emissions
are much smaller in magnitude than for the highest ranked categories, the broadly-based
wastewater category consistently ranks in the top ten emitters for nitrous oxide and methane
emissions in the U.S. NACWA's review focused on emissions from POTWs, which are a
fraction of the total wastewater treatment category emissions.
The emissions from POTWs in the 2012 Inventory are essentially the same as those in the 2011
Inventory, with some clarifications added to the text. NACWA's comments on the 2011
Inventory requested that all values used in the equation to calculate emissions be provided to
enable the calculations to be easily reproduced. NACWA appreciates the response to this request
with the addition of Table 8-15, which provides the values for the variables used in calculating
the nitrous oxide emissions for 2012 and previous years.
Comment: NACWA agrees with the additions made to the Planned Improvements section and
encourages EPA to investigate additional data sources as soon as possible. Since the 2008 Clean
Watershed Needs Survey (CWNS) is not detailed enough to be used in the Inventory and the
2004 CWNS data is likely outdated, additional data sources are necessary to ensure the accuracy
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
17
-------
of future Inventories. NACWA supports EPA's investigation of the data available at
www.biogasdata.org and from ongoing research in the U.S. and abroad. However, NACWA also
urges caution in using results from studies that were not designed to produce nationally-
applicable results. Relying on studies that are not representative of utilities nationwide may
actually increase the uncertainty of the estimates. NACWA agrees that EPA's plan to review
inventories from other countries for additional data and methodologies may be useful, as long as
any information used is directly applicable to wastewater treatment processes in the U.S.
Comment: As NACWA has explained in comments on the Inventory in previous years, the
Association believes that the nitrogen loading rates for N20EFFLUENT are sourced incorrectly
and that using information from the existing National Pollution Discharge Elimination System
(NPDES) database will yield more accurate and justifiable loading rates. The NPDES permitting
program represents long-term, nationwide facility performance which would allow emissions
estimate projections over the time series represented in the Inventory. If EPA decides not to
investigate its own databases, the average nitrogen loading rate of 15.1 g N/capita-dayl
represents the industry standard and is supported by a wealth of data widely confirmed in U.S.
practice, as explained in our previous comments and supported by data collected by NACWA
from 48 U.S. POTWs. This result represents all domestic sources of nitrogen, the use of other
nitrogen-containing compounds, and both residential and commercial sources.
Comment: Outside of the Wastewater Treatment section, the Inventory's Executive Summary
and Introduction should state more clearly that the Inventory's purpose is for information, not
regulation. EPA should ensure that all of its offices understand the purpose of the Inventory and
recognize that the Inventory's industry-wide methodologies are largely inadequate for facility
level emissions, such as those required by EPA's Greenhouse Gas Reporting Rule and the Clean
Air Act Title V and Prevention of Significant Deterioration (PSD) permitting programs.
Commenter: David McCabe
Clean Air Task Force (CATF)
Comment: Methane from Petroleum and Natural Gas Systems:
In our January comments on the Expert draft of the inventory, we raised a number of issues that
we summarize here. Although EPA has noted most of the issues we raised in the discussion text
of the public draft inventory ("Draft Inventory"), the emissions estimates in that version have not
been substantially modified from the expert draft inventory. Consequently, the inaccuracies we
identified remain in the inventory estimates. As such, we re-confirm our January comments,
which we have attached to this document for your convenience, with some updated figures, and
have made additional specific suggestions about how EPA might handle identified inaccuracies
in the draft inventories here.
We raised three principle issues in the January comments: Emissions from completion of oil
wells with hydraulic fracturing (HF), emissions from completion of gas wells with HF, and
emissions from pneumatic controllers (PCs).
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
18
-------
In addition we raise two additional issues in these comments: Methane from venting of
associated gas from oil wells (during production), and general comments about the approach
taken to tabulating emissions in the Inventory.
Comment: Emissions from completion of oil wells with hydraulic fracturing. EPA continues to
use a very low emissions factor for oil well completion - 733 scf per completion - that pre-dates
the use of HF on oil wells. Since most oil wells are now hydraulically fractured, and the use of
HF substantially increases potential emissions per well (just as for natural gas wells), the use of
this ancient emissions factor for all oil well completions substantially underestimates actual
emissions.
We reference the recent comments from Environmental Defense Fund (EDF) on the Draft
Inventory. EDF reviewed data from a number of sources that show that both potential and actual
emissions from oil well completions after hydraulic fracturing can be hundreds of times greater
than the 733 scf per well completion EPA uses in the Draft Inventory. EDF recommends that
EPA use analysis of EPA's Greenhouse Gas Reporting Program (GHGRP) data for reported
emissions from well completions in oil-bearing formations. This analysis finds that reported
actual cumulative emissions (including wells where gas was vented, wells where gas was flared,
and wells where gas was captured into pipelines, during flowback) were an average of 6.2 metric
tons of methane per completion or recompletion, based on reports on 1,754 completions and
recompletions. We support EDF's recommendation that EPA use this data to revise the estimates
for well completion of oil wells for wells that use hydraulic fracturing, EDF's suggestion that oil
well completion emissions be reported with sub-categories for wells with and without hydraulic
fracturing, and EDF's suggested approach for estimating the number of oil wells that use
hydraulic fracturing.
Estimating methane emissions from oil well completions in this manner would clearly be more
accurate than EPA's current method. EPA must promptly address this rather manifest inaccuracy
in the final 2014 inventory. If EPA is unable to provide a more accurate estimate of emissions
from oil wells completions in the final 2014 inventory, a statement directly noting this issue is
warranted. We suggest adding the following to page 3-55 (suggested additions in bold):
-line 11:"... increase again with the widespread use of hydraulic fracturing in tight
formations."
- After the period on line 13. "Note that the inventory methodology has not been updated
to reflect emissions during well completion or re-completion after hydraulic fracturing,
and thus the inventory likely underestimates emissions from this source."
Comment: EPA has revised the methodology for estimating emissions of methane from
completion of gas wells. As in our January comments, we generally support this revision, as the
revised data appears to be based on more robust data and the result is much more transparent.
However, as we noted in our January comments, EPA's methodology is flawed because it fails to
account for the significant fraction of gas well completions at facilities that do not report data to
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
19
-------
the GHGRP. Thus, the activity data reported in table A-126 of the inventory is clearly an
underestimate of actual activity. As we suggested earlier, EPA should use data from state
databases or commercial databases, compared to reports to GHGRP, to calculate the fraction of
wells that are owned by firms that report data to the GHGRP, and then adjust the activity data
from GHGRP using this fraction, to get a more accurate estimate of the number of completions
occurring nationwide. The suggested approach is described in somewhat more detail in our
January comments, as EPA mentions in the Public Draft Inventory (p 3-71, lines 1-3).
Although adjusting the GHGRP to a proper estimate of national activity is not trivial, the current
figure is clearly an underestimate of national emissions and we believe EPA would set a poor
precedent by using unadjusted data in the Inventory. If EPA chooses not to adjust the GHGRP
data, as suggested or by some other approach, EPA should acknowledge in the inventory that the
issue exists. Currently, in the Public Draft, this issue is not mentioned directly, but rather is
indirectly referred to (in response to CATF comments) under "Well Counts and Completion and
Workover Counts" under "Planned Improvements." It should be raised with a statement to the
effect of, "This methodology undercounts emissions from completions and workovers with
hydraulic fracturing, to the extent that it undercounts completion and workover events, because
not all well facilities report emissions and activity data to GHGRP." This statement belongs in
either the completions text in the QA/QC section, or the completions text in the Recalculations
section. Additionally, it would be a great example to list under Uncertainty and Time Series
Consistency. For example, starting on line 35 of p 3-66,
The IPCC guidance notes that in using this method, "some uncertainties that are not
addressed by statistical means may exist, including those arising from omissions or
double counting, or other conceptual errors, or from incomplete understanding of the
processes that may lead to inaccuracies in estimates developed from models." An
example would be the probable undercount of completion and workover events with
hydraulic fracturing (see below). As a result, the understanding of the uncertainty of
emissions estimates for this category will evolve and will improve as the underlying
methodologies and datasets improve.
Comment: Pneumatic Controller emissions:
As described in our January comments, GHGRP data shows significantly higher emissions from
pneumatic controllers (PCs) than the Draft Inventory reports. Since the GHGRP uses emissions
factors derived from EPA/GRI (1996), as does the Inventory, the apparent difference between
the two is in device counts. Although the GHGRP clearly undercounts devices (by not capturing
all wellpads, or any emissions from gathering), it must be more accurate than the current activity
figures used in the Inventory. As noted in our January comments, since not all wellpads report
under the GHGRP, it shows that total emissions from oil and gas wellpads were, at a minimum,
861 Gg methane in 2012. The Public Draft reports emissions from both Gas Production and Oil
Production of 692 Gg methane, less than was reported in the Expert Draft (787 Gg methane), so
this gap has widened significantly. While we recognize that updated activity data for 2012 may
increase the figures in the Inventory, relative to the Public Draft, we anticipate that the gap
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
20
-------
between the Inventory figure and the GHGRP figure will remain: for 2011, the Public Draft
Inventory reports emissions of 752 Gg methane from oil and gas production, while the GHGRP
data shows 835 Gg methane. Again, this is troubling because the GHGRP only covers a portion
of the facilities that the Inventory is intended to cover yet its emissions figures are higher than
the Draft Inventory.
As mentioned in our January comments, the gap would be much larger if the more accurate
emissions factors from the Allen et. al (2013) study were used in place of the GHGRP emissions
factors. The data available today suggests that the Allen et al. emissions factors are the best
available today. As described in our January comments, correcting the GHGRP with the Allen et
al. emissions factors produces a national methane emissions estimate of 1,140 Gg, even without
scaling emissions up to reflect production facilities that do not report to GHGRP.
Thus it appears that both the activity data and emissions factors used in the inventory for PCs are
not the most accurate data available. If EPA cannot use the more accurate, recent data we have
suggested here, EPA should continue to note that data (as is done in the Public Draft) and
commit to examining this data in the coming year.
Comment: Venting of Associated Gas from Oil Wells:
It appears that the Inventory underestimates venting from oil wells by a substantial amount. To
our understanding, venting of associated gas from oil wells during production (i.e., casinghead
gas venting) is listed in the inventory as "Stripper Wells" under Vented Emissions in Petroleum
Production, and is listed as 14.2 Gg methane for 2012. Last year's inventory listed the same
value for 2011.
GHGRP data shows much higher emissions of methane from "Associated Gas Venting and
Flaring." For 2011, 175 Gg methane emissions were reported to GHGRP; for 2012, the figure
was 90 Gg. Some of this is due to emissions of methane from flares, due to incomplete
combustion in the flame. This portion of the methane emissions can be accurately estimated, by
comparing C02 emissions from associated gas venting and flaring to methane emissions from
that source. As described below, CATF analyzed the GHGRP data in this way, finding that 60%
to 90% of the GHGRP methane emissions from associated gas venting and flaring are due to
venting, and thus the 14.2 Gg methane figure in the Draft Inventory is significantly too low.
We compared the emissions of C02 and CH4 reported from each facility reporting "Associated
Gas Venting and Flaring" emissions to the GHGRP, for both 2011 and 2012. The GHGRP uses a
default factor of 2% for emissions of methane from flares, due to incomplete combustion (40
CFR Part 98.233(n)(l), Eq. W-19). Using this factor, we subtracted away the maximum methane
that could be due to incomplete combustion in flares from each individual facility report. To be
conservative, we also considered a case where the factor for incomplete combustion for methane
in flares was 5%, in case some facilities used this higher factor to calculate their emissions.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
21
-------
In either case, many facilities have methane emissions well in excess of that due to incomplete
combustion, and this "excess methane," which is due to venting of associated gas, is significant
at the national level. In 2011, vented emissions, calculated with the 2% emissions factor, were
157.5 Gg methane; in 2012 they were 60.1 Gg CH4. Raising the incomplete combustion factor to
5% drops these figures somewhat, to 154.0 Gg CH4 and 56.3 Gg CH4, respectively. This
analysis of the GHGRP data shows that venting of associated gas from oil wells is much higher
than the 14 Gg reported in the Draft Inventory.
Comment: General Comments on the Organization of the Inventory for Oil and Gas:
As new data emerges on emissions from oil and gas facilities, it is very important that EPA use
that data in the most appropriate way in the inventory. In some cases it may not be best for EPA
to force the data into the framework used in the inventory, particularly where a) oil and natural
gas production are separated and b) natural gas production is broken down, for almost every
individual source, into NEMS regions.
It is appropriate that EPA has not developed emissions factors for each NEMS region for well
completion and workover with HF, and it greatly improves the clarity of the inventory that these
emissions are presented separately in Table A-126. We believe that EPA should have taken the
same approach last year when updating the methodology for estimating liquids unloading (LU)
emissions. The report submitted by API and ANGA on LU emissions did not recommend
developing distinct emissions factors for each NEMS region for wells that vent during LU with
and without plunger lifts. Instead, API and ANGA concluded it was more appropriate to estimate
national emissions by applying their entire dataset to national activity drivers. EPA, in
calculating emissions for each NEMS region, concludes that national LU emissions were
substantially lower than API and ANGA concluded. Moreover, the calculated emissions factors
for LU wells vary tremendously between NEMS regions that are not designed to capture
differences in geology, age of wells, or anything else that might affect LU emissions. The NEMS
region emissions factors are simply not credible.
EPA should use the national emissions factor approach used for completion / workover
emissions for LU.
Additionally, when data from the GHGRP is superior to other available data, EPA should use
that data, even if it does not readily allow separation of emissions between the oil production and
natural gas production sectors. As EPA has recognized in, for example, GHGRP Subpart W and
NSPS Subpart OOOO, these sectors are really one industry, and the distinction between the two
is necessarily arbitrary. At present the Inventory reports that over 60% of emissions from PCs are
from oil production, so it may be more appropriate to simply list PC emissions under oil
production, with the "included elsewhere" designation for PCs under gas production.
Finally, we comment here on the Draft Inventory's discussion of Methane Measurement Studies
(p. 3-71). First, we note that the Brandt et al. study mentioned in this section is quite specific that
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
22
-------
emissions from oil and natural gas, specifically, were underestimated in the 2013 Inventory
(which reported higher emissions than the Draft 2014 Inventory. Quoting from Brandt et al.:
"We find ... measurements at all scales show that official inventories consistently underestimate
actual CH4 emissions, with the NG and oil sectors as important contributors..(emphasis
added). The title of Brandt et al.'s paper is "Methane Leaks from North American Natural Gas
Systems." The Draft Inventory's over-generalization of Brandt et al.'s findings must be
corrected.
EPA asks for input on how information from top-down ambient studies should be used to update
the inventory. One non-quantitative way that EPA should use this information is to put the
bottom-up inventory in context. The bottom-up inventory is essential for understanding the
specifics of GHG emissions so that mitigation priorities, for example, can be examined.
However, the bottom-up inventory clearly does not capture all emissions from oil and gas
operations. Scientifically, it is not clear that the emissions from the sector as a whole are most
accurately measured with the bottom-up measurements available to date. As such, it may be
appropriate to calculate the leak rate from the entire industry (for life-cycle analysis, for
example) using different methodologies. Separately, as ambient studies continue and techniques
are developed, they will illuminate sources that must be reexamined in bottom-up studies. For
these reasons, the top-down methodologies are strongly complementary to the bottom-up
approach.
It would be appropriate for EPA to mention, in this section, the value of top-down studies in
providing independent data on overall emissions from the industry, and on identifying specific
potential issues in the inventory.
Comment: Emissions Data for Wood Biomass Combustion:
CATF requests that EPA provide a clearer explanation of the data on C02 emissions from wood
consumption reported in section 3.10 of the Draft Inventory, particularly the data that are
reported for electricity generation units (EGUs). EPA describes the approach it used to determine
the amount of C02e emitted by EGUs that combust woody biomass in the Methodology passage
at 3-79 of the Draft Inventory, but it is difficult—if not impossible—to replicate the results that
EPA achieved using the database referenced by the Agency.
According to CATF's understanding of the Methodology passage at 3-79, EPA has determined
that the Acid Rain Program's "bottom-up" data for woody biomass consumption by EGUs are
better than the EIA Monthly Energy Review data for those same sources. EPA made the same
determination in the 2013 US GHG Inventory of Emissions and Sinks, but neither the 2013
Inventory nor the 2014 Draft Inventory explains the Agency's preference for the Acid Raid data.
(Id.; 2013 Inventory at 3-79). The lack of an explanation is particularly problematic because,
notwithstanding its concerns about the EIA data for woody biomass consumption by EGUs, EPA
considers EIA's national estimate for total woody biomass consumption to be accurate. (2014
Draft Inventory at 3-79). In any event, EGU biomass consumption data for 2012 is lower in the
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
23
-------
Acid Rain dataset than it is in the EIA dataset. So in order to use both the Acid Rain data for
woody biomass consumption by EGUs and the EIA data for total national consumption, EPA
adjusts the consumption/emissions data for the other three sectors (Industrial, Residential,
Commercial). (Id). It appears that that EPA calculated the difference between the EIA data and
the Acid Rain data for woody biomass consumption by EGUs and then proportionally allocated
that difference to those other sectors' EIA scores.
Assuming that is in fact how EPA determined the woody biomass consumption and emissions
data reported in Tables 3-55 and 3-56, CATF was unable to reproduce the reported results for the
Electricity Generation sector using the ampd.epa.gov database — which in turn prevented us
from assessing the reported results for the other three sectors. We note, however, that the
reported EGU emissions for 2012 (21.0 Tg C02e) are an order of magnitude higher than they
were in 2008 (2.8 Tg C02e), 2009 (2.4 Tg C02e), 2010 (2.6 Tg C02e), and 2011 (2.4 Tg
C02e). (Draft 2014 Inventory at 3-78, Table 3-55). We also note that emissions from EGU
consumption of woody biomass reported during earlier years (e.g. 2008-2011) appear to be too
low when compared to emissions data that CATF received in response to queries at
ampd.epa.gov. Finally, CATF notes that Acid Rain database appears to be significantly under-
inclusive. For example, when CATF queried the database for national CO2 emissions from EGUs
that combust "wood" and "other solid fuel," the result was comprised of emissions from only a
handful of facilities located in just five states.
Comment: Discussion and Presentation of Global Warming Potentials (GWPs) from IPCC's
Fifth Assessment Report in the introduction and Annex 6.1:
In the Draft Inventory, EPA has presented the GWPs from IPCC's Fifth Assessment Report
(AR5) incompletely. We commend EPA for committing to using the GWPs from the 2007
Fourth Assessment Report (AR4) in next year's inventory, in compliance with UNFCCC
guidelines. However, the more recent AR5 GWPs are now considered more accurate, and it is
important that EPA let readers know about these updates. The material presented in the
Introduction, and in Annex 6.1, does not accurately report what AR5 reports for GWPs, and the
problem is particularly acute for methane from "fossil" sources such as coal, oil, and natural gas.
For all GHGs, AR5 reports two GWPs. For one, the climate carbon feedback ("cc-fb") effects
are included when the radiative forcing from the target gas (the non-C02 GHG) is calculated; for
the other GWP, the cc-fb are not included in this calculation. However, GWPs are calculated
relative to the radiative forcing caused by C02, and the cc-fb is included for the calculation of
radiative forcing from C02 in all GWP calculations. That is, when the GWP for methane is
calculated "without the cc-fb," the radiative forcing for methane without the cc-fb is compared to
the radiative forcing for C02 with the cc-fb. For this reason, IPCC states that it is likely that the
GWPs with the cc-fb included are more accurate. (See page 731 of AR5). As such, the Draft
Inventory, which only presents AR5 GWPs without the cc-fb, (Draft Inventory at 1-9, Box 1-2)
is not presenting the most accurate information to readers.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
24
-------
Furthermore, specific to methane, EPA should also include the GWPs that IPCC calculates for
fossil methane. The table on page 1-9 omits this. There is no reason not to include the GWP for
fossil methane in the table. While the issue was not properly handled in earlier editions of IPCC
Assessment reports, it is a simple matter: C02 from the oxidation of fossil methane is additional
C02 in the climate system, whereas C02 from oxidation of biogenic methane is not. EPA must
report the best available scientific information. Consistency with earlier, less accurate IPCC
reports, is not an acceptable reason to leave this information out of the Table. After all, it is just
an informational table. However, many users will read this table to read that AR5 says the GWP
of CH4 is 28, period. The notes at the bottom of the table are not sufficient.
In summary, a line should be added for fossil methane. It would have no entries for SAR, TAR,
or AR4, so it would be clear that it is new, and that there is no analogue in the previous ARs.
Secondly, an additional column with the AR5 GWPs with the cc-fb included is needed.
Likewise, Annex 6.1 is quite helpful, but it appears to not even acknowledge the GWPs in AR5
calculated with the cc-fb included, nor the separate GWPs for fossil methane. Thus, it is not
accurately using the recommendations of AR5.
For example, natural gas, petroleum, coal mines & abandoned coal mines, stationary & mobile
combustion, petrochemical and iron/coke production together account for 43% of US methane
emissions. Thus, if using the GWPs without the cc-fb included, the correct change to methane
emissions (in C02e) for AR5, relative to AR4 (table A280), would be (5 * 0.43 + 3 * 0.57) / 25
or 15.4%, not the 12% reported in table A280. As mentioned above, AR5 says that it is likely
that the values with the cc-fb included are more accurate, so the more accurate GWPs are
actually 34 for biogenic methane and 36 for fossil methane. Therefore the most accurate value
for the change to methane emissions (in C02e) for AR5, relative to AR4 (table A280) would be
(11 * 0.43 + 9 * 0.57) / 25 or 39%. That's a significant difference, and ignoring all of these other
values for GWP does a real disservice to readers of this section.
Therefore, tables A276, A280, A281 should be updated to use the fossil methane GWP for those
sources, and to discuss and show the differences if the GWPs with cc-fb included (for all non-
CO2 GHGs) are included.
Comment: Minor Suggested Corrections / Clarifications (by page number)
Page ES-14, Line 13: Strike "observed." The drop in emissions is inferred, largely from changes
in activity drivers.
Page 1-6: Suggest the following addition (in bold):
"Tropospheric ozone is produced from complex chemical reactions of volatile organic
compounds and/or methane mixing with NOx in the presence of sunlight."
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
25
-------
Page 1-9 Box 1-2: In the GWP discussion in the ES, the importance of forcing by secondary
products of primary forcers is mentioned. Updates to the calculations of forcing by secondary
products is an important factor in the changes in GWPs in AR5. Suggest the following addition
(in bold):
"In the AR5, the IPCC has applied an improved calculation of C02 radiative forcing and
an improved C02 response function in presenting updated GWP values. IPCC also
applied updated calculations of indirect radiative forcing for some gases. Additionally,
the atmospheric lifetimes of some gases have been recalculated, and updated background
concentrations were used. In addition, the values for radiative forcing and lifetimes have
been recalculated for a variety of halocarbons, which were not presented in the SAR."
Page 1-10 Very end of section 1.1, Add this sentence (copied from ES):
"The use of IPCC AR4 GWP values in future year inventories will apply across the entire
time series of the inventory (i.e., from 1990 to 2013) in next year's report."
Page 3-70, line 1: the correct table reference is Annex Table A-135
Page 3-71, line 16: The Brandt et al study is not listed in the References section. Also, please
make an effort to provide a URL, whenever possible, to all of the documents listed in the
References section, particularly EPA documents.
Tables A125-A130 reference a number of documents not listed in the references section on pp
A200-A202. EPA should attempt to get as many of the memos and other documents listed as
references onto the website, and provide hyperlinks to those documents in the references section!
Table A125: the EF for Liquids Unloading w/o plunger lifts for region MC is messed up (it is
written as 190,17 scfy/well, so either a digit is missing or the comma is in the wrong place.
There are no references listed for the petroleum section. (And a lot of other sections. Maybe
those works are cited at the end of Annex 3?)
Commenter: Darren Smith
Devon Energy
Comment: Due to our position as an early-adopter of reducing emissions from production
processes, Devon holds unique knowledge about the processes involved and the physical
phenomena that shape emissions for hydraulically fractured wells. It was this expertise and
knowledge - and the resulting discovery that EPA's previous estimates for methane emissions
from the flowback of hydraulically fractured wells were heavily inflated - that led Devon to take
an active role in encouraging EPA to refine the previously adopted emission factor for
hydraulically fractured wells. It is this same expertise that leads Devon now to commend EPA
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
26
-------
for the proposed changes in the 2012 Draft GHG Inventory. The proposed changes to this year's
inventory have brought the inventory one step closer to an accurate and complete emissions
profile for the natural gas production sector.
Comment: Devon supports the use of the Greenhouse Gas Reporting Program for this emission
factor, replacing the previous estimate based on Natural Gas Star Data, which was not fit for the
purpose of establishing emission factors. Devon has provided numerous comments criticizing the
previous emission factor for methane emissions from the flowback of hydraulically fractured
wells. The crux of the criticism was that Natural Gas Star recovered volumes were used as a
proxy for emissions from vented well completions. Natural Gas Star data is not fit for emission
factor determination. This new method drastically improves accuracy of the factor, and can form
the basis for making adjustments to the inventory in the future, as industry technology continues
to reduce emissions in the oil and gas sector. While there is still room for improvement due to
the GHGRP's use of the choke flow calculation methodology, the change represents a significant
improvement in accuracy.
More importantly, through the adoption of net emission factors, EPA provides a framework by
which future greenhouse gas reporting rule results can be used to continually refine the emission
factors for methane from hydraulically fractured well completions. As the greenhouse gas
reporting rule further refines its calculation and reporting methodologies, and as industry
improves its technology and practices to further reduce emissions, the emission factors for
methane from well completions can and should be adjusted accordingly. Given that net
emissions factors will closely mimic GHGRP data, updates to the emission factor can be easily
automated, so that an accurate emissions profile can be captured each year. Finally, creating net
emission factors that more closely match the GHGRP data will provide the public confidence in
the accuracy of this particular data program.
This potential, and the ability for the public to verify greenhouse gas reporting program data,
provides transparency to the method by which the factors are determined. This allows policy
makers and the public to better understand the different emission profiles for different equipment
configurations, and for the federal and state governments to make policy decisions based on
accurate data.
Commenter: Erica Bowman
America's Natural Gas Alliance
Comment: ANGA appreciates the changes EPA has made in developing the 2014 Draft GHG
Inventory, which incorporates new data sources and methodologies that more accurately reflect
actual emissions. These changes include the establishment of technology-specific emissions
factors for wells with hydraulically fractured completions and workovers. We encourage EPA to
continue upgrading the GHG Inventory with net emission factors in place of potential emission
factors as more data become available. We would also support further sub-categorization to
recognize the differences between hydraulically fractured completions and hydraulically
fractured workovers.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
27
-------
Comment: For the past several years, ANGA has submitted comments on EPA's Draft GHG
Inventories. Included in those comments were concerns that EPA has overestimated emissions
from natural gas production activities, particularly emissions associated with the liquids
unloading, and well completions and workovers. In the 2013 GHG Inventory, EPA adjusted the
methodologies for estimating the frequency of well re-fracturing and emissions from liquids
unloading. These changes contributed to a reduction in estimated 2010 Field Production
emissions from Natural Gas Systems of 54 percent. ANGA supported these changes, which more
accurately accounted for actual field practices.
In the 2014 Draft GHG Inventory, EPA adjusts the methodology for completions and workovers
with hydraulic fracturing. These adjustments establish four technology-specific emissions factors
for wells with hydraulically fractured completions and workovers: (1) hydraulic fracturing
completions and workovers that vent; (2) hydraulic fracturing completions and workovers that
flare; (3) hydraulic fracturing completions and workovers with Reduced Emission Completions
(RECs); and (4) hydraulic fracturing completions and workovers with RECs that flare. These
emissions factors are based on data submitted to EPA under the 20 II and 2012 Greenhouse Gas
Reporting Program (GHGRP) Subpart W. Compared to data used in the 2013 GHG Inventory,
the GHGRP data shows a higher percentage of hydraulically fractured well completions and
workovers using RECs, a higher percentage of hydraulically fractured well completions and
workovers that flare, and fewer emissions per hydraulically fractured completion and workover
that vented. We believe that the adjustment to the emissions factor for hydraulically fractured
well completions and workovers that vent is closer to representing actual emissions. The
GHGRP data used by EPA support ANGA's long-held contention that EPA's estimate that 9,000
thousand cubic feet (Mcf) of natural gas is released per uncontrolled well completion is flawed
due to its reliance on data from the Natural Gas STAR program.
Comment: Although the new emission factors for uncontrolled well completions better represent
actual industry practices, they remain higher than measured results from the recent study by
researchers at the University of Texas-Austin and supported by Environmental Defense Fund
(UT Austin/EDF study). At 41 metric tons (MT) methane per vented well completion, for
example, the estimate in the Draft 2014 GHG Inventory is within one order of magnitude of the
range found for similarly configured completions in the UT Austin/EDF Study, which found a
range of 0.5-4 MT methane per completion event for those wells vented directly to atmosphere.
Much of this difference can be attributed to the choke flow calculation methodology option in
the GHGRP. The choke flow calculation methodology was not designed for use in multi-phase
flow applications, and as such can often deliver erroneous results when compared to direct
measurement. ANGA encourages EPA to remove outlier data from the emission factor
calculation and use only measured data in the GHGRP for the calculation of emission factors, not
data derived from the choke flow equation methodology.
Comment: As noted above, ANGA supports the use of GHGRP data to establish emission
factors and strongly believes that EPA should continue using this data source to refine the
emission factors for hydraulically fractured well completions and workovers. As industry
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
28
-------
technology and practices improve to further reduce methane emissions and the GHGRP
continues to update its calculation and reporting methodologies, the emission factors for
hydraulically fractured wells and completions should be adjusted accordingly. In addition to
improving the accuracy of the GHG Inventory which is a common goal of both EPA and the
natural gas industry, creating emissions factors that more closely match the GHGRP data will
provide public confidence in and increase uniformity across EPA's data programs.
While ANGA continues to believe that EPA's estimate of the number of uncontrolled well
completions and workovers is too high, we understand that this number will decrease
significantly in future years as the 2015 and later GHG Inventories will factor in the REC and
completion combustion device requirements included in the Oil & Gas New Source Pollution
Standards. This rule requires the use of RECs for almost all completions and workovers after
January 1, 2015 and required flowback emissions to be routed to a completion combustion
device starting in October 2012. As a result, the 2015 GHG Inventory, which reports estimated
emissions from 2013, should have significantly lower emissions from these activities.
Comment: In response to EPA's request for input on the assumptions regarding the historic use
of RECs, we support the recommendations made by Devon in its comments on the expert review
draft and public review draft of the 2014 GHG Inventory. As EPA considers other changes to the
inventory, we would support sub-categorization of pneumatic controllers to high bleed, low
bleed, and intermittent categories and the use of appropriate actual emission factors for each
category using GHGRP data, the UT Austin/EDF study, and other recent and upcoming studies.
Comment: Given the magnitude of the changes that the Agency has made over the past four
years both increasing and decreasing estimated emissions from natural gas production, the
underlying data and assumptions must be rigorous and well supported. ANGA appreciates the
changes EPA has made to its methodology for estimating emissions from liquids unloading, its
estimate of the frequency of work overs, and its methodology for hydraulically fractured well
completions and workovers. We encourage EPA to continue updating its methodology and
emissions factors with technology- and region-specific emissions factors based on valid data,
assumptions and calculations. However, given the underlying uncertainties of the current data,
ANGA does not support the use of the emissions estimates presented in the GHG Inventory as
the basis for any analysis or regulatory action.
Commenter: Karin Ritter
American Petroleum Institute
Comment: General:
API supports the changes made to the 2012 U.S. GHG Inventory including the advances made in
updating the national emission estimation methodology and increased use of site specific
industry data that is becoming available through the Greenhouse Gas Reporting Program
(GHGRP). When accounting for these changes the resulting non-combustion emissions from
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
29
-------
Natural Gas Systems are estimated to be 162.3 million metric tonnes of C02e (C02 - 35.2; CH4
- 127.1). This represents a 1.07% of natural gas withdrawals for 2012. API encourages EPA to
state this clearly early in the discussion on Natural Gas Systems to enhance understanding of the
data by potential users.
Comment: General:
For this Public Review of the draft 2012 national inventory, API is providing comments
regarding emission estimation for Petroleum Systems and Natural Gas Systems. Our comments
reiterate some of the discussions on recalculations that were part of the U.S. GHG Inventory
expert review phase and also point out areas for future collaboration where EPA is planning
future improvements. API supports further review and analysis of the GHGRP data with the
overarching goal of ensuring the quality and validity of data being used for deriving new national
emission factors.
In addition, results from on-going GHG emission studies are expected to be published this year,
and API is willing to continue its collaboration with EPA to incorporate relevant new
information in the 2012 U.S. GHG Inventory and beyond.
Comment: General:
API supports the continued disaggregation of emission source information and, if applicable,
emission reductions, to provide better transparency for "net" emissions for each source type. The
approach historically used by EPA of lumping together reduction activities for multiple
inventory sources made it difficult to attribute these reductions to specific inventory source
categories. Emission reductions reported for "Other Production", "Other Processing", "Other
Transmission" and "Other Distribution" in Table A-135 are larger than those shown in the
Expert Review Draft and provide less transparency about the sources of these emission
reductions.
Comment: General:
Where appropriate for the source category, API supports the continued use of data reported
through the GHGRP and other relevant "bottoms-up" studies to develop "net" emission factors
for specific source categories. API also recommends that EPA recalculate "net" emission factors
for relevant source categories on an annual basis, using the GHGRP data and any relevant new
"bottoms-up" studies, for each successive inventory in order to reflect changes in emissions due
to expanded regulatory and voluntary reductions. This allows EPA to highlight, in the U.S. GHG
Inventory, changing operating practices due to regulatory requirements being phased in by the
petroleum and natural gas sector over the next few years.
Comment: General:
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
30
-------
API advises EPA to carefully analyze and screen GHGRP reported data to identify data outliers
and enable verification and/or correction or exclusion of suspect data entries and prevent the use
of incorrect data in the derivation of emission factors (EFs). As discussed previously with EPA,
during the Expert Review phase, the GHGRP data may potentially include incomplete or
incorrect data due to ambiguity in implementation of approved EPA procedures, errors in
applying the GHGRP calculations, faults in data aggregation and reporting, and partial reliance
on Best Available Monitoring Methods (BAMM). Despite these discussions and detailed
analysis provided to EPA to highlight the impact of erroneous data and outliers it seems that
EPA did not modify their calculations published in the Public Review version of the 2012
inventory now under consideration.
Comment: Petroleum Systems Emissions:
Page 3-54 and Page 3-55. Editorial Comment:
API has noted a redundancy in the text presented in rows 28-32 of page 3.54 with rows 10-13 of
page 3.55.
Comment: Petroleum Systems Emissions:
Page 3-59. Recalculation Discussion: Accounting for Voluntary Emission Reductions:
Under its recalculation discussion EPA seeks comment on its update to the Petroleum Systems
section to include Natural Gas Star reduction data. EPA has added an accounting for voluntary
emission reductions to the CH4 emissions from Petroleum Systems, and it indicates that this is
from reassigning reductions that were previously included under the Natural Gas Systems (as
referenced on page 3-70).
API supports this change but notes that the reductions attributed to the Petroleum Systems lacks
the level of transparency that was previously provided for Natural Gas Systems. To address this,
API recommends that Section 3.6 for Petroleum Systems in the annex should include a table that
is equivalent to Tables A-135 and A-136 in the Natural Gas Systems.
Comment: Petroleum Systems Emissions:
Page 3-59. Planned Improvements Oil Well Completions and Workovers:
EPA is discussing its planned improvement to the U.S. GHG Inventory for oil production to
allow for differentiation between completions with and without hydraulic fracturing. EPA is
seeking comments on the topic as part of its future improvements effort since comments they
received during the Expert Review phase indicate that 75-90% of all new oil wells are completed
with hydraulic fracturing. Some commenters suggested that updated emission factors could be
developed using data from recent studies and EPA is quoting a wide range of potential average
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
31
-------
emission factors that are being considered without providing any explanations or justifications
for these emission factors:
• 6.2 Mg CH4 (GHGRP based on gas well completions and workovers in Oil formations
for wells with and without control);
• 3.1 Mg CH4 (UT Austin/EDF; wells with controls);
• 9.7 and 24.7 Mg CH4 (Wattenberg and Eagle Ford data, wells without control)
API wants to emphasize that existing data from recent field studies or from extrapolation from
gas wells in oil formations do not provide a reliable representation of potential emissions from
oil well completions and workovers. API is willing to work with EPA to assess data that may be
used for future improvements of the emission factors used to characterize this emission source.
Comment: Petroleum Systems Emissions
As an additional item for future improvement, EPA is repeating its requests from the Expert
Review draft for data on the Oil wells refracture rate, which EPA currently assumes to be 7.5%
per year. As previously stated API concurs that field data for Oil well completions with and
without hydraulic fracturing is currently sparse. However, EPA's assumption of a 7.5%
workover (or refracture) rate for all oil wells seems higher than is expected based on industry's
experience.
API is willing to work with EPA to develop a reasonable oil well refracture rate for potential use
in future inventories.
Comment: Natural Gas Systems Emissions
Page 3-69. Recalculation Discussion: Gas Wells Completions and Workovers: Alternative
Approach to Emission Factors Categories:
During the expert review phase of the U.S. GHG Inventory API supported EPA's derivation of
new Emission Factors for gas wells completions and workovers utilizing GHGRP data. API has
also noted the need for careful screening of reported data to make sure that erroneous entries and
outliers are not used in these calculations.
Moreover, API has recommended that EPA collapse the proposed four categories for grouping
gas well completions and workovers with hydraulic fracturing into only two categories.
Therefore, in response to EPA's request for comments during the Public Review phase of the
inventory, API reiterates its previous comments and maintains that the future relevance of the
four distinct operating practices for which EFs were derived ought to be reconsidered. Newly
proposed changes to estimating and reporting emissions for flowback events for hydraulically
fractured completions and workovers 1 and the phasing in of compliance with the Oil and Natural
Gas (NSPS)2 will likely result in few to no events without reduced emissions completion
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
32
-------
(RECs), and those with RECs will generally include both venting and flaring for short periods of
time.
As described before, API is proposing an alternate two-category approach that may be adopted
for future inventories and which would entail the derivation of emission factors that are
representative of completions and workovers with hydraulic fracturing and limit significant
changes in subsequent inventories in view of the expected operational changes:
• Non-REC Completions and Workovers (Vented only); and
• REC Completions and Workovers (vented and flared).
Table 1 below reintroduces a modified version of the EFs from such an alternative approach, as
provided by API during the expert review phase of the U.S. GHG Inventory. The results are
presented for both the 2011 and 2012 GHGRP data (with outliers removed) and these two
categories are expected to provide a good characterization of emissions from these emission
sources and will enable tracking industry's transition to the use of reduced emission completions
and workovers. Based on discussions with EPA it became clear that EPA's count of vented
completions and workovers without RECs includes completions with zero emissions. API's
initial calculation approach excluded these data sets for the non-REC completions and
workovers. This has been revised in API's modified analysis shown in Tablel. For 2012, 466
non-REC vented completions and 95 non-REC workovers were reported with zero emissions.
Year
Category
Total CH4
Emissions,
tonnes C02e
# events
Tonnes
CH4/event
Scf
CH4/event
# data
sets
2012
Data
Non-REC
Completions and
Workovers (Vented
only)
1,121,164
3,037
17.58
915,596
252
REC Completions
and
Workovers (vented
and flared)
219,364
3,051
4.21
269,854
333
2011
Data
Non-REC
Completions and
Workovers (Vented
only)
2,803,608
2,957
45.15
2,351,503
346
REC Completions
and
Workovers (vented
and flared)
430,161
4,815
4.25
221,572
319
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
33
-------
Comment: Natural Gas Systems Emissions
Page 70. Planned Improvements: Completion and Workover Counts:
In its discussion about future Planned Improvements, EPA is addressing issues that were brought
up repeatedly in API's discussions with EPA. API has provided comments before about the
inconsistency in accounting for the total number of completions and workovers, due to
ambiguous language in the GHG Reporting Program.
At EPA's request, API has surveyed its members and summarizes below the findings that point
out the differences between EPA and API's completions and workover counts, which impact the
calculated emission factors.
• EPA assumed the number of completions is equal to the sum of total completions
reported and completions with purposely designed separating equipment (RECs). API
assumed the RECs were a subset of the total completions reported. This was confirmed
by seven (7) member companies.
• EPA assumed the number of workovers is equal to the sum of vented workovers, flared
workovers, and REC workovers. API assumed the total number of workovers was equal
to the sum of the vented and flared workovers, and that workovers with purposely
designed separation equipment were a subset of this total. This was confirmed by five (5)
member companies.
• Where data sets provided a count of workovers with REC, but no count of vented or
flared workovers and zero emissions, EPA assigned these as vented workovers with REC.
API treated these as invalid data sets. For 2012, this applied to 11 data sets, representing
21 workover events. The API analysis has been revised to include these data sets, as
reflected in Table 1 above.
Comment: Natural Gas Systems Emissions
Page 3-71. Planned Improvement: Methane Measurement Studies:
EPA is requesting feedback on how measurements from top-down studies can be used to update
its emissions estimates. As API stated before, studies such as Petron 2012 and Miller et al. 2013
focus on inverse flux modeling which employs emission concentration data from aircrafts,
ground-based or towers over a regional area or on ambient hydrocarbon species ratios analysis.
These studies have either been regional and do not fully represent natural gas production in the
US (e.g. Petron 2012), or do not represent current operations (Miller 2013 and Petron 2012).
Additionally, these studies are a "snapshot" in time and do not necessarily give any indication of
emission rates over a longer time period such as annual. It is well know that bottoms-up methods
like Allen et al. have much better accuracy over top-down methods. Since EPA's greenhouse gas
inventory, uses a bottoms-up approach in itself, especially for quantifying CH4 emissions, it is
more appropriate to use other bottoms-up approaches as data sources and for inventory
verification.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
34
-------
There are several studies underway that attempt to combine top-down and bottoms-up methods
to better understand and reconcile the differences. Until such time, EPA should only consider
studies that measure emissions directly from the individual sources or activities.
Additionally, API wants to reiterate that no top-down study will be able to produce granular level
information provided by the EPA inventory with respect to individual sources or activities within
a sector. At best, these top-down methods can be used for gross verification of the inventory
estimates.
Comment: Comments on Appendix A, Table A-135:
EPA has revised the voluntary emission reduction data in the table. The "other production"
category increased from 40 Gg CH4 in the expert review version to 619.3 Gg in this version.
There is no explanation of the change in reductions other than EPA reallocated some from the
natural gas systems to petroleum systems.
API contends that this change is a step backward in the transparency of the emission reduction
data and urges EPA to elaborate on how the change was calculated and what it includes. This
does not apply only to the onshore production segment since the same increase is noted in the
"other" reductions for the other industry segments listed in Table A-135.
Comment: Comments on Appendix A, Table A-141:
There seems to be an error in Table A-141. API's recalculation of the production sector
emissions indicates that the value shown for condensate tanks in this table (2252 Gg) is not the
net emissions. The net emissions for this source should be 164.9 Gg CH4.
Comment: Comments on Appendix A, Table A-143:
EPA revised the emission estimate for C02 from flares. While in the Expert Review version
9,868.6 Gg C02 were reported (Table A-141) in this version we note a value of 12,738.8 Gg
(Table A-143). This appears to combine flaring from production and processing operations. API
is requesting that EPA explain this new value and state specifically what industry segment it
represents, or break out emissions associated with production operations separately from
processing.
Comment: General Editorial:
API suggests that EPA keep the same order for the emission sources in the tables presented for
each industry sector. This would certainly help when reviewing tables side by side. For example,
EPA has moved the location of the emissions for gas well workovers among the different tables.
In Table A-125, these emissions are presented with completions and well drilling, while in Table
A 1-43, workover emissions are presented separately after tanks.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
35
-------
In summary, API appreciates the opportunity to provide comments during the public review
phase of the 2012 U.S. GHG Inventory. EPA noted some errors and omissions that need to be
addressed prior to finalizing the inventory while reiterating comments provided during the
Expert Review phase and indicating areas for potential future improvements.
Commenter: David Isaacs
Semiconductor Industry Association
Comment: The current IPCC guidelines were established in 2006 with data collected in 2004 and
earlier. While the data used to devise these methods represented the best available data at the
time, it no longer represents the most accurate data available. The default emission factors
contained in the current IPCC guidelines were based on 75 emissions characterization data sets,
which may no longer be representative of the processes and equipment used throughout the
industry.
Comment: In 2013 EPA issued a final rule governing the reporting of greenhouse gas emissions
for the semiconductor industry in the United States, codified at 40 C.F.R. Part 98 Subpart I. As
part of the development of this regulation, SLA member companies, several process equipment
manufacturers, and SEMATECH, contributed to a large data collection effort resulting in a
substantial amount of new data. The participants in this data collection undertook an extensive
effort to characterize the processes deployed in our industry. The data collected was from
equipment processing different wafer sizes and multiple semiconductor companies and
equipment suppliers. It includes every fluorinated greenhouse gas currently used in
semiconductor plasma etch processing and chemical vapor deposition chamber cleaning. The
new data brings the total number of data sets to 1182.
SLA believes that the additional data used in the development of Subpart I will result in more
accurate and more representative reporting of PFC emissions from semiconductor fabs in the
United States as compared with the current IPCC guidelines used internationally to report
emissions from our industry. EPA evidently concurs with this conclusion through the adoption
of the regulation. Therefore, in order to improve the reporting of emissions globally and ensure
consistency in reporting methods, SLA requests that EPA work to update the current IPCC
guidelines to reflect this new data. Updating the IPCC guidelines will improve the consistency
of the data contained in the U.S. inventory with the information available globally, and also
improve the accuracy of the global data. SLA would be pleased to assist EPA in this endeavor.
Commenter: David Lyon
Environmental Defense Fund
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
36
-------
Comment: Environmental Defense Fund (EDF) previously submitted comments on the Draft
Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2012 (Draft 2014 Inventory)
during the expert review comment period. As stated in our previous comments, we recommend
that EPA updates the Petroleum Systems source category to account for methane emissions from
co-producing well completions with hydraulic fracturing. EDF has prepared a white paper on
methane emissions from co-producing well completions that summarizes our analyses of several
recent data sources including the Greenhouse Gas Reporting Program Subpart W, Allen et al.
2013, and initial production data from the Eagle Ford, Bakken, and Wattenberg fields. Based on
these analyses, we estimate that 2012 methane emissions from co-producing well completions
are between 96 and 247 Gg CH4, comparable to the Draft 2014 Inventory estimate of 217 Gg
CH4 emissions from hydraulically-fractured gas well completions and workovers.
Commenter: Nathan Matthews
Sierra Club
Comment: The Sierra Club files these comments on the February 2014 draft 1990-2012
Greenhouse Gas Inventory. We offer the following concerns:
• For gas production, although EPA proposes to revise sector wide emissions estimates
downward, recent science based on atmospheric measurements indicates that a strong
upward revision is appropriate.
• The "UT Austin EDF" Study provides further indication that the inventory's estimate
of gas systems emissions is too low. Emissions from pneumatic controllers, in
particular, are likely to be underestimated.
• The draft inventory does not include emissions from unconventional (e.g.,
hydraulically fractured) petroleum wells.
• EPA's outdated figure for methane's global warming potential is far lower than recent
estimates.
Comment: Atmospheric Studies Indicate That Gas Systems Have Far Higher Emissions:
The February 2014 draft reduces EPA's estimate of total emissions from gas production. Yet
several recent published studies based on regional atmospheric methane measurement indicate
that estimates EPA proposes to lower were already too low.
We briefly summarize these atmospheric studies here. The first group of studies looked at
particular regions. Two studies led by researchers with the National Ocean and Atmospheric
Administration (NOAA) Earth System Research Laboratory that have directly measured methane
in the atmosphere in other regions have estimated high leak rates. The first of these studies
explains that by monitoring methane, propane, benzene, and other volatile organic compounds
in the air around oil and gas fields, the authors can estimate oil and gas production's
contributions to these pollutant levels. According to the study authors, their "analysis suggests
that the emissions of the species we measure are most likely underestimated in [1990-2010]
inventories," perhaps by as much as a factor of two, which would imply a leak rate of about
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
37
-------
4.8% of production. A second announced NOAA study suggests that leak rates in the Uinta may
be as high as 9%. Additionally, we note that a California study identified a 17% leak rate for oil
and gas (presumably primarily oil) operations in the Los Angeles basin.
The second group of studies, released in the last four months, looks at nationwide gas production
emissions and specifically criticizes the prior inventory as underestimating GHG emissions from
gas production. In December of 2013, a paper published by Scot M. Miller et al. in the
Proceedings of the Natural Academy of Sciences reviewed atmospheric measurements of
methane and concluded that "The US EPA recently decreased its CH4 emissions factors for
fossil fuel extraction and processing by 25-30%) (for 1990-2011), but we find that CH4 data
from across North America instead indicate the need for a larger adjustment of the opposite
sign." In other words, rather than reducing the estimated leak rate from 2.4% to something
approaching 1.5%, EPA should have increased its estimate to at least 3%. In February, a paper
published in Science similarly concluded that the then current inventory underestimated
methane emissions from gas production—indicating that the February 2014 draft is a change in
the wrong direction.
Sierra Club has not identified the likely reason for the discrepancy between these "top down"
assessments incorporating atmospheric measurements and EPA's "bottom up" estimate based on
individual components, practices, and emission factors. Assuming the atmospheric studies to be
correct, factors contributing to this discrepancy may include underestimation of the number of
wells, a system wide underestimation of per component emission factors, drastic
underestimation of emissions from particular sources (perhaps pneumatics or liquids unloading),
or there may be some other cause. Although we are unable to recommend a particular correction
to the inventory fully reconcile the inventory with these studies, we strongly encourage EPA to
devote attention to this issue.
Comment: The "UT Austin-EDF" Study Further Indicates That The Inventory Underestimates
Gas Systems Emissions:
The 2014 draft acknowledges a study by David Allen, of University of Texas, Austin, et al. and
sponsored by the Environmental Defense Fund (EDF) as a source of additional information
regarding gas and petroleum system emissions; the draft generally refers to this work as the "UT
Austin EDF study." This study also generally indicates that the inventory underestimates
emissions from gas systems. For the wells and completions included in this study, observed
emissions were similar to average gas system emissions implied by the 2013 GHG inventory.
However, the UT Austin EDF study found much higher utilization of reduced emission
completions than are contemplated by the EPA inventories, resulting in drastically lower
emissions from that particular slice of the lifecycle. These reduced completion emissions were
offset, however, by increases from other components, such as pneumatics, in excess of those
assumed by the inventory. These observed high rates of emissions from activities other than
completions should be expected to apply industry wide, indicating that 2013 inventory
underestimated these emissions. More generally, the UT Austin-EDF study should be assumed to
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
38
-------
represent the top end of performers, insofar as the wells included were from large industry
players who opted in to the study and who had notice that measurements would be taken. EPA
should look critically at emission estimates that would indicate that the industry as a whole
performs better than the subset of players and wells included in the UT Austin-EDF study.
Comment: The Inventory Likely Underestimates Emissions from Pneumatic Devices:
As noted above, the UT Austin-EDF study indicates that the inventory significantly
underestimates emissions from pneumatic devices. Until individual pneumatic devices are
reported pursuant to Subpart W, EPA should adopt an approach such as the one recommended
by EDF in their separate comment on the February 2014 draft.
Comment: Petroleum Systems Estimates Must Account for Unconventional Production:
There can be no disputing that hydraulic fracturing has changed the face of American petroleum
production and has been employed in a large percentage of petroleum wells for a number of
years. FracFocus, the national hydraulic fracturing chemical registry managed by the Ground
Water Protection Council and Interstate Oil and Gas Compact Commission, includes records
from 12,056 oil wells that were hydraulically fractured in 2012. Yet the 2014 draft of the
inventory estimates petroleum system emissions only using emission factors for conventional
production. As explained in comments submitted separately by the Environmental Defense Fund,
many of the tools EPA proposes to use to estimate gas systems emissions can also be applied to
petroleum systems. Although these tools are imperfect, they can provide a much more accurate
estimate of emissions than the draft inventory's inaccurate assumption that hydraulic fracturing
is not used in petroleum wells.
Comment: EPA Uses an Outdated, and Far Too Low, Estimate of Methane's Global Warming
Potential:
The inventory discusses methane's global warming potential (GWP) on the 100 year timeframe,
and estimates this potential as 21. EPA explains that it uses this value pursuant to UNFCCC
reporting obligations. Id. Yet as EPA recognizes, this value does not represent the best available
science. As an interim measure, EPA provides an annex with many charts explaining the impact
of using the 2007 Intergovernmental Panel on Climate Change (IPCC) 100 year methane GWP
estimate of 25, Annex 6.1, but even that estimate has been superseded in the intervening seven
years of research. Most importantly, the IPCC's Fifth Assessment Report estimates an aggregate
100-year methane GWP of 34, and an even higher estimate of 36 for methane emitted from fossil
sources.
EPA must therefore take available steps to encourage this reporting obligation to be updated to
reflect the best available science. These steps include including informing other federal entities
participating in negotiation of these agreements of the importance of using recent science. As an
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
39
-------
interim measure, EPA should present an annex using the methane GWP data from the IPCC AR5
report, as the draft inventory does for the IPCC AR4 data.
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse Gas
Emissions and Sinks: 1990-2012
40
-------
Supplemental Material Received
Appendix A
Environmental Defense Fund White Paper on Methane Emissions from Co-Producing Well Completions
Appendix B
Energy Innovation Missing Methane Issue Brief
Appendix C
Damascus Citizens for Sustainability Extended Report on a Preliminary Investigation of Ground-Level
Ambient Methane Levels in Manhattan, New York City, New York
Appendix D
Damascus Citizens for Sustainability citing Phillips et al. 2012
Appendix E
Damascus Citizens for Sustainability citing Pieschl et al. 2013
Appendix F
Damascus Citizens for Sustainability Report on a Survey of Ground-Level Ambient Methane Levels in
the Vicinity of Wyalusing, Bradford County, Pennsylvania
Appendix G
Damascus Citizens for Sustainability citing Report to the Clean Air Council of the June 8, 2012, on
Field Inspection and Methane Sampling Survey
Appendix H
Damascus Citizens
Appendix I
Damascus Citizens
Appendix J
Damascus Citizens
for Sustainability citing Tollefson 2013
for Sustainability citing Tollefson 2012
for Sustainability citing Miller et. al 2013
Comments Received during the Public Review Period on the Inventory of U.S. Greenhouse
Gas Emissions and Sinks: 1990-2012
20
-------
Appendix A
-------
Co-Producing Wells as a Major Source of Methane Emissions:
A Review of Recent Analyses
Prepared by Environmental Defense Fund
March 2014
The Environmental Protection Agency's ("EPA's") New Source Performance Standards
("NSPS") for the oil and natural gas sector require that hydraulically fractured natural gas wells
reduce theircompletion emissions using eitherreduced emission completions ("RECs") or
flaring.1 EPA defines a "gas well" or "natural gas well" as "an onshore well drilled principally for
production of natural gas"2 and, depending on how this definition is interpreted, a number of
wells that co-produce oil (or other liquids) and natural gas ("co-producing wells") may not need
to control theiremissions underthe REC requirements in the NSPS.
Many completions of these co-producing wells, however, produce substantial pollution
that can be cost-effectively mitigated using the same clean air measures that have effectively
reduced emissions from hydraulically fractured gas wells. Extending clean air protections to co-
producing wells is vital given recent trends within the oil and gas industry. Over the last two
years, rising oil prices and low natural gas prices have caused new drilling activity to
increasingly shift to shale formations rich in oil and condensates. Reflecting this trend, the U.S.
Energy Information's ("ElA's") most recent Annual Energy Outlook predicts that domesticoil
production will grow significantly through 2020, driven primarily by increases in tight oil
production (see Figure 1).
Figure 1. US Petroleum and Other Liquids Supply, 1970-2040 (EIA)
History 2012 Projections
1970 1680 1680 2000 2010 2020 2030 2040
1 With limited exceptions, all fractured and refractured natural gas wells will be required to use RECs as of January
1,2015. 77 Fed. Reg. 49,490, 49,497 (Aug. 16,2012).
240C.F.R. §60.5430.
1
-------
This analysis synthesizes available information on per-completion emissions factors, the
cost-effectiveness of mitigating those emissions using RECs or high-efficiency flaring, and,
where possible, the total amount of methane that would be reduced by deploying these
completion protections at co-producing wells. Table 1 synthesizes data from the following
sources:
• A February, 2014 Stanford/Novim Study in the journal Science entitled "Methane
Leakagefrom North American Natural GasSystems;" ("Stanford/Novim Analysis")3
• ICF International's Report from March, 2014 entitled "EconomicAnalysis of Methane
Emissions Reduction Opportunitiesinthe U.S. Onshore Oil and Natural Gas
Industries;" ("ICF Report")4
• A2013 analysis in the Proceedings of the National Academy of Sciences led by the
University of Texas entitled "Measurements of methane emissions at natural gas
production sites in the United States;"5 ("UT Study")
• EDF's analysis of the oil and natural gas portion of EPA's Greenhouse Gas Reporting
Program ("EDF Subpart W Analysis");6 and
• An analysis completed by EDF and Stratus Consulting of well completion reports in
the Bakken, Eagle Ford, and Wattenberg field ("EDF/Stratus Analysis").
These sources all indicate that co-producing well completions are a substantial source of
methane emissions, with total estimated emissions much largerthan the figure reported in
EPA's official inventory ofgreenhousegasemissions. EPA'scurrentemissionfactorforco-
producing wells derives from a 1996 study of conventional oil wells, and very likely
underestimates emissions from the hydraulic fracturing techniques that are prevalent today.
3 A.R. Brandt et al., Methane Leaks from North American Natural Gas Systems, 343 Science 733 (Feb. 14, 2014),
available at http://www.novim.orq/imaqes/pdf/ScienceMethane.02.14.14.pdf.
4 The report is available at http://www.edf.org/sites/default/files/methane cost curve report.pdf.
5 David T. Allen et al., Measurements of methane emissions at natural gas production sites in the United States,
PNAS Early Edition (2013), available at www.pnas.orq/cqi/doi/10.1073/pnas. 1304880110.
6 EDF, Comments on "Draft Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2012" (included in the
supplemental information for this analysis).
2
-------
TABLE 1: Summary of Co-producing Emissions, Cost-Effectiveness, and Mitigation Potential
Data
Potential
National
REC Cost
Flaring Cost
Methane
Sources
Emission
Emissions
Effectiveness
Effectiveness
Mitigation
Factor (MT
Estimates***
($/MT CH4)
($/MT CH4)
Potential
CH4)
(MTCH4)
without
savings
with
savings
(MTCH4)
Stanford/
40.2'
120,000****
778
92
114,000
No vim
Analysis*
ICF Report
6.6**
96,000
n/a
n/a
96.57
94,000
UTStudy*
193.5
153.8
-132.7s
19.19
n/a
EDF Subpart
21.8
163,000
1,435
170
140,000
W Analysis
EDF/Stratus
15.7
247,000
3,578
3,314
424
235,000
Analysis
*Analysis includes potential emissions factor only. Cost-effectiveness and mitigation potential
derived using common assumptions described below.
**This EF includes both vented emissions controlled emissions so is not a true potential
emissions factor.
*** Estimates provided by the authors of each individual study.
**** This estimate only reflects emissions from three major production basins, and therefore
understates total national emissions.
The remainder of this white paper provides additional information on the development
of an emission factor for co-producing wells, the cost-effectiveness of mitigating these
emissions, and overall methane mitigation potentials.
Potential Emission Factor
The above-described analyses determine potential emissionsfactorsforco-producing
well completions using several different methods, including direct measurement, analysis of
Subpart Wdata, and analysis of initial oil and gas production. All of these analysesfind
potential emissions are significantly greaterthan the emissions factorforoil well completions
currently in EPA's annual greenhouse gas inventory (0.0141 tons of methane percompletion).
Given that EPA's currentemissionsfactorisdated and was based on emissionsfrom
completions of conventional, non-hydraulically fractured wells, the more recent studies
described below suggest that the official inventory is likely underestimating the extensive
methane emissionsfrom co-producing well completions. Moreover, neitherthe current NSPS
7 Weighted average of emission factors forwells in the Bakken, Eagle Ford, and Permian Basins.
8 On average, these wells would achieve net savings of $25,630 by selling gas recovered during completions,
assuming $4/Mcf.
3
-------
nor the regulations of most states require control of completion emissions from co-producing
wells.9
UT Study. The UT Study measured various large sources of methane in the production
sector, including 27 well completions in various geographic areas across the country. Six of the
measured completions were at co-producing wells that produced significant amounts of
hydrocarbon liquids,10 and, for each of these completions, researchers directly measured
potential and actual methane emissions. Actual completion emissions from these co-producing
wells ranged from 1.7 to 5.0 metric tons ("MT") CH4, though all of the wells controlled
completion flowback emissions with eitherflaring or a combination of RECs and flaring. The UT
study estimated potential emissions as the total volume of gas vented, flared, and sent to sales
from initiation of flowback until the reported completion end time. The potential emissions
from these wells, which would be more indicative of uncontrolled completions, ranged from
81.9 to 414.4 MT ChU, with an average value of 193.5 MT of ChU/completion.11
Table 1. Measured and potential emissions of co-producing wells from Allen, etal. (2013)
Completion
Event
Emission
Controls
Measured
Emissions
(scfCFU)
Potential
Emissions
(scfCFU)
Measured
Emissions
(MTCH4)
Potential
Emissions
(MTCH4)
GC-1
Flaring
105,000
5,005,000
2.0
96.4
GC-2
Flaring
90,000
4,250,000
1.7
81.9
GC-3
REC&
Flaring
260,000
21,500,000
5.0
414.1
GC-4
REC&
Flaring
180,000
13,000,000
3.5
250.4
GC-6
Flaring
247,000
12,200,000
4.8
235.0
GC-7
Flaring
90,000
4,320,000
1.7
83.2
Average
162,000
10,030,000
3.1
193.5
Subpart W Analyses. EDF also evaluated completion data from 2011 and 2012 that was
reported to EPA under its greenhouse gas reporting rule for oil and gas systems (known as
"Subpart W").12 Subpart W does not require reporting of oil well completion and workover
9 Notably, Colorado does require that co-producing wells perform reduced emission completions. Co. Oil & Gas
Conserv. Comm'n ("COGCC") Rule 805(b)(3)(A).
10
David T. Allen et al., Measurements of methane emissions at natural gas production sites in the United States,
PNAS Early Edition (2013), available at www.pnas.orq/cqi/doi/10.1073/pnas. 1304880110. See also EDF, Analysis of
Co-Producing Well Completions (updated Mar. 2013) (included in the supplemental information forthis analysis).
11 EDF, Analysis of Co-Producing Well Completions (Dec. 2013). The underlying study analyzed a total of 26 well
completions.
12 EDF, Comments on "Draft Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2012" (included in the
supplemental information for this analysis).
4
-------
emissions. Nonetheless, in 2011 and 2012 there were 1,754 reports of completions and
workovers from wells in formations classified under Subpart W as "oil formations." EDF
performed a separate analysis of Dl Desktop data to assess if these completions were actually
oil wells.13 In approximately 75% of the counties from which these completion reports came,
over half of the wells with first production in 2011 & 2012 were oil wells. Using the same
approach that EPA used to estimate emission factors for completionsfrom the entire GHGRP
dataset, EDF has derived emission factors forthis subset of wells located in oil formations
(Table 3). The average emission factorfor all oil formation completion and workovers is 6.2 MT
ChU/event, or more than 400 times higher than the current oil well completion emission factor.
EDF also developed separate emission factorsforeach combination of emission controls
reported under Subpart W: uncontrolled ("vented") completions, completions controlled with a
flare, completions controlled with a REC, and completions controlled with both flares and REC.
The emission factors for the four categories range from 3.1 MT ChU/event for completions with
RECto21.8MT ChU/eventforvented completions.
The ICF Report also uses Subpart Wdata to develop an emission factorfor hydraulically
fractured oil wells. From this data, the Report develops an emissions factor of 344,000 scf
ChU/completion or6.6 MT ChU/completion, which is an average value including both controlled
and uncontrolled completions.
Table 3. Oil well completion and workover emission factors developed from 2011 &2012
GHGRP Subpart Woil formation type sub-basins using the same method as EPA for
developing the natural gas completion and workover emission factors
Category
Completions
(# events)
Workovers
(# events)
Completions
& Workovers
(# events)
Completions
EF (MT
ChU/event)
Workovers
EF (MT
ChU/event)
Completion
& Workover
EF (MT
CH_/event)
Vent
320
147
467
21.8
7.6
17.3
Flare
221
66
287
3.7
2.5
3.4
REC
186
0
186
3.1
N/A
3.1
REC+Flare
17
0
17
11.7
N/A
11.7
Ambiguous
708
89
797
1.5
0.0
1.3
All events
1,452
302
1,754
6.6
4.2
6.2
Initial Production Analyses. The Stanford/Novim Analysis evaluated 2,969 well
completions in the Bakken, Eagle Ford, and Permian basinsfor2011 using the Drillinglnfo HPDI
Database.14 The analysis estimated potential emissions from these tight oil wells by converting
13 Data obtained from Drillinglnfo, Dl Desktop, http://info.drillinqinfo.com/products/di-desktop/.
14 A.R. Brandt et al., Methane Leaks from North American Natural Gas Systems, 343 Science 733 (Feb. 14, 2014),
available at http://www.novim.orq/imaqes/pdf/ScienceMethane.02.14.14.pdf. The relevant data is contained in
the supporting documentation forthe study
(http://www.sciencemaq.orq/content/suppl/2014/02/12/343.6172.733.DC1/Brandt.SM.datafile.xlsx).
5
-------
peak gas production to a daily initial production rate. It then assumed that production during
flowback increased linearly with time for 9 days prior to initial production and all such methane
emissions were vented, or understood differently, that completion emissions correspond to 4.5
days of initial gas production.15 Using this methodology, the analysis determined potential
emissionsfactorsforthe Bakken (31.1 MTChU/completion), Eagle Ford (90.9 MT
ChU/completion), and Permian (31.2 MT ChU/completion) Basins.
The EDF/Stratus analysis takes a similar approach, using initial production values to
understand potential completion emissions at co-producing wells. Stratus Consulting initially
performed an analysis of 100 well completions in the Bakken, assuming a 7 to 10 day
completion event with gas production increasing from zero to the initial production value in a
non-linear fashion over the course of the completion. Accordingly, Stratus assumed that total
gas production over the 7-10 day completion event would equal 3 average days of gas
production.16 As with the Stanford/Novim analysis, Stratus assumed all of this gas was vented.
EDF subsequently extended this analysis to approximately 9,500 wells in the Bakken,
Eagle Ford, and Wattenberg fields.17 Only oil wells were analyzed for the Eagle Ford and
Wattenberg fields; North Dakota does not distinguish between oil and gas wells so all Bakken
wells were assumed to be oil wells. Across all wells, the analysis found an average potential
emissions factor of 15.7 MT ChU/completions with averages of 18.0, 24.7, and 9.5 MT
ChU/completion in the Bakken, Eagle Ford, and Wattenberg respectively.
Cost Effectiveness
Other than the ICF Report, none of the above non-EDF analyses calculated the cost-
effectiveness of controlling completion emissions using RECs or high-efficiency flaring.
Accordingly, we applied consistent cost assumptions to all of the analyses above, except the ICF
Report. For RECs, we assumed 95% control efficiency and used EPA's cost of performing a
reduced emission completion ($29,713)18 to calculate cost-effectiveness. Across all studies, we
calculated a REC cost-effectiveness without a credit for captured gas ranging from $154 -
$3,578/MT ChU reduced. Using production data from approximately 9,500 wells in the Bakken,
Eagle Ford, and Wattenberg fields, we calculated a REC cost-effectiveness with credit for gas
15 This methodology is set forth in Francis O'Sullivan & Sergey Paltsev, Shale gas production: potential versus actual
greenhouse gas emissions, Envtl. Res. Letters 7(4):044030 (Nov. 26, 2012).
16 Memorandum from Leland Deck, Stratus Consulting, to Peter Zalzal and Vickie Patton, Environmental Defense
Fund, re: Methods Memo on VOC Cost-Effectiveness in Controlling Bakken Shale Combined Oil and Gas Wells
During Well Completion (Mar. 30, 2012) (included in the supplemental information for this analysis).
17 EDF, Spreadsheets analyzing Bakken, Eagle Ford and Wattenberg wells (included in the supplemental
information for this analysis).
18 EPA, Oil and Natural Gas Sector: Standards of Performance for Crude Oil and Natural Gas Production,
Transmission, and Distribution, Background Technical Support Document for Proposed Standards (July 2011),
available at http://www.epa.aov/airaualitv/oilandgas/pdfs/20110728tsd.pdf.
6
-------
capture. With a credit for gas savings (based on an assumed gas price of $4.00/Mcf), we
calculated a median cost-effectiveness of $3,314/MT CH4 reduced and also calculated cost-
effectiveness for the top 25% and top 10% of wells, as shown in the table below.
Table 4. E DF / Stratus REC Cost-Effectiveness for Median and Top 25 and 10 Percent of Wells
Percentile
REC Cost
Effectiveness with
gas capture credit
($/MT CH4)
Mitigation
Potential
(MTCH4)
Mitigation
Potential (%
of total)
10%
$544
60,643
40.9
25%
$1,266
97,430
65.7
50%
$3,314
126,508
85.3
To calculate flaring cost effectiveness, we assumed 95% destruction and removal
efficiency ("DRE") and multiplied this by the emission factor to get flaring emission reductions.
We then divided the EPA cost estimate of flaring completion emissions from a well ($3,523) by
the flaring emission reductions for each of the analyses.19 Across all studies (excluding the ICF
Report) we calculated a flaring cost-effectiveness ranging from $19 - $424/MT CH4 reduced.
The ICF Report includes its own cost assumptions about performing high-efficiency
flaring, which are substantially higher than those in EPA's NSPS. ICF assumes flaring has a 98
percent control efficiency and a capital cost of $50,000, with an additional $6,000 in fuel costs
for ignition. ICF estimates the cost-effectiveness of flaring to be $1.86/Mcf of methane
($97/MT CH4) for completion gas. The ICF report did not examine the cost-effectiveness of
RECs for co-producing wells.
Mitigation Potential
Determining inventory-wide mitigation potential requires scaling up emissions
nationally and then applying percentage reductions associated with mitigation technologies.
TheStanford/Novim Analysis, the ICF Report, the EDFSubpartWAnalysis, and the EDF/Stratus
Analysis all provide national estimates of emissions from co-producing wells, which we describe
in greater detail below. The UT Study does not scale these specific emissions nationally and we
have not provided a separate scale up of those emissions here.
7
-------
• Stanford/Novim Analysis. The Stanford/Novim analysisfound that co-producing well
completions accounted forapproximately 120,000 MT CH4 in 201 1.20 The analysis
assumed all emissions were vented and multiplied emissionsfactors in the Bakken,
Eagle Ford, and Permian Basins by the total numberof completions in those basins.
Because the 120,000 MT CH4figure includes only emissionsfrom these three basins, it is
not a true national figure.
• ICF Report. ICF used itsemissionsfactorof 344,000 scf ChU/completion (6.6 MT
CH4/completion)from Subpart W along with the most recent API Quarterly Completions
Report showing 15,382 hydraulically fractured oil well completionsfor2011. Using these
values, ICF calculated completion emissionsof 5 Bcf CH4or96,000 MTCH4.
• EDF Subpart WAnalysis. EDF applied emissionsfactors we calculated from Subpart W
to the 2012 Draft GHG Inventory activity data of 15,753 oil well completions.21 This
resulted in emission estimates between 49,000 MT CH4 (assuming all RECs) and 343,000
MT CH4 (assuming all emissions vented), or 182,000 MT ChUiftheuse of emission
controls among the 15,753 oil well completions is assumed to be distributed in the same
way as the Subpart Wdataset. Because some wells are already controlled, we assumed
the national proportion of uncontrolled completions was 43%, the same as the Subpart
W dataset, and applied the emission factorfor vented completions. We use this 147,000
MTCH4 value for purposes of determining mitigation potential.
• EDF/StratusAnalysis. TheEDF/Stratusanalysisdidnotisolatehydraulicallyfractured
wells, but instead derived an average emission factor applicable to all co-producing well
completions. Accordingly, EDF applied emissionsfactorswe calculated using the Stratus
methodology to EPA's 2012 Draft GHG Inventory activity data of 15,753 oil well
completionsforan emissions estimate of approximately 247,000 MT CH4 annually.
Translating these national emissions estimates into mitigation potential requires
applying control efficiencies. The ICF Report assumes flaring achieves 98% DRE, and accordingly
suggests mitigating completion emissions from co-producing wells could achieve 94,000 MT
CH4 in annual reductions.
on
A.R. Brandt etal., Supplementary Materials for Methane Leaks from North American Natural Gas Systems 30,
343 Science 733 (Feb. 14,2014), available at
http://www.sciencemaa.org/content/suppl/2014/02/12/343.6172.733.DC1/1247045.Brandt.SM.pdf.
21 Although not all oil wells completions use hydraulic fracturing, FracFocus, the national hydraulicfracturing
chemical registry managed by the Ground Water Protection Council and Interstate Oil and Gas Compact
Commission, includes records from 12,056 oil wells that were hydraulically fractured in 2012. Reporting to
FracFocus is voluntary in many states, which implies that the actual number of hydraulically fractured oil wells is
higher than 12,056. Accordingly, we have used the draft inventory activity data as a reasonable proxy for the total
number of hydraulically fractured oil well completions.
8
-------
The Stanford/Novim analysis does not calculate mitigation potential, and so, consistent
with the two EDF analyses, we conservatively assume flaring or gas capture achieves a 95%
control efficiency. Because both the Stanford/Novim analysis and EDF/Stratus analysis assume
all emissions are vented, we apply the 95% control figure directly to total emissions estimates,
resulting in annual mitigation potentials of 114,000 MT ChU and 228,000 MT ChU respectively.
Because EDF's Subpart W analysis assumes some wells are already controlled, we apply the
95% control effectiveness only to the subset of emissions that are vented for an annual
mitigation potential of 140,000 MT Chk
Conclusions
Although neither EPA regulations nor the regulations of most states require control of
emissions from co-producing well completions, these emissions are a potentially significant
source of methane and other harmful pollutants. Recent studies and analyses - drawing from a
variety of data sources including field studies of well completions, Subpart W reports, and well
completion databases-suggestthat emissionsfrom an uncontrolled co-producing well
completion range from 15.7 MT of CH4 to nearly 200 MT. At a national level, these emission
factors suggest total co-producing well completion emissions between approximately 96,000 to
247,000 MT, comparable to emissionsfrom natural gas well completions (209,000 MT CH4 in
the latest EPA annual inventory). Current control technologies for natural gas well completions
-including RECs where gathering infrastructure is available, and high-efficiency flaring in other
situations- can be readily applied to co-producing well completions. This white paper suggests
that applying those technologies to co-producing well completions would yield emission
reductions on the orderof 94,000 to 228,000 MT peryear, or2.63 to 6.38 million MT C02-e
(using 100-year GWP of 28).
9
-------
Appendix B
-------
ENERGY INNOVATION ¥
POLICY g TECHNOLOGY, LLC
The Mystery of the Missing Methane
Advances in the scientific understanding of methane emissions highlight
the need for improvements to the EPA emissions inventory
By Chris Busch
25 March 2014
1, Summary
The U.S. Environmental Protection Agency (EPA) recently released a draft of its 1990-2012 greenhouse
gas (GHG) emissions inventory. While the EPA is in many ways at the frontier of global best practice, the
agency needs to take action to account for the accumulating evidence that the GHG inventory is
omitting a significant fraction of methane emissions, the second most prevalent contributor to climate
change. The new draft inventory estimates that emissions fell almost two percent in 2012 compared to
2011, and it revises downward previous estimates of methane emissions for the natural gas sector. For
example, 2011 emissions are almost 10 percent lower in the 2014 draft inventory than they were in the
2013 inventory. These downward revisions are being made despite increasing scientific evidence that
the EPA should be increasing its estimate of emissions.
Just one week before the draft inventory was released, the journal Science published a landmark study
(Brandt et al.. 2014) that concludes that the EPA inventory is undercounting emissions by a significant
margin. The study brings together, for the first time, the full body of existing evidence on methane
leakage. It estimates that there are 7-21 teragrams (Tg; 1012 grams) of methane missing from the EPA
inventory and concludes that some of this methane is likely coming from the natural gas system. This
quantity, 7-21 Tg, is equivalent to roughly 25-75 percent of the total methane emissions in the
inventory and is two to four times the EPA's current estimate of methane emissions from the natural gas
system.
The EPA needs to develop a plan to collect and analyze real-world data to narrow the uncertainty ranges
and provide a better understanding of methane emissions, especially from the natural gas system. New
technologies for detection and measurement of methane emissions can help the EPA achieve this goal.
Additional resources should be dedicated to this objective.
1
-------
2. Bottom-up vs. top-down studies of methane emissions
The EPA emission inventory relies on "bottom-up" studies of methane emissions. Bottom-up studies
involve component-level sampling on the ground, at the source. The EPA uses the results from these
studies to calculate emission factors for different activities that make up the natural gas system,
including production, processing, transmission, and distribution. These emission factors—essentially,
typical levels of emissions per unit output for different components of the system—are applied to
natural gas production activities to calculate activity-specific emissions, and then are summed to
estimate total system-wide emissions. As the EPA inventory for the natural gas system is constructed,
uncontrolled emissions are first estimated using the process above (the "potential emissions"), then
regulatory initiatives and voluntary information provided by companies are taken into account to
produce estimated emissions.
Figure 1. Methane emissions are invisible to the naked eye
Methane emissions from this storage tank are visible not the naked eye but an infrared
lens reveals their existence. Photo source: New York Times.
One of challenges with bottom-up studies is that they require the participation of landowners and
natural gas companies. Researchers must obtain permission in order to enter a property and directly
measure emissions, and have not found it easy to do this. There is some reason to believe that the
producers that have voluntarily participated are the cleanest, lowest-emitting operators. This, in
combination with the great heterogeneity in types of operations and geology across gas-bearing basins,
means that it is difficult for bottom-up studies to collect data from a broad enough array of sources for
the sampling to be representative.
"Top-down" studies are a second, distinct approach for measuring methane emissions. These studies
are based on atmospheric sampling from aircraft or tall towers. Top-down studies provide great
accuracy with respect to the quantity of total emissions (though some uncertainty is introduced by
wind-blown methane that might enter or exit the study area before being sampled). Traditionally, the
weakness of top-down studies has been the difficulty of discerning the contribution of different sources
the overall level observed level of methane. Many top-down studies have not even attempted to
attribute the methane sampled in the atmosphere to particular sources on the ground. However,
emerging techniques are making progress in allowing identification of likely sources for atmospherically
sampled methane.
2
-------
3. The missing methane
Brandt et al.'s paper is innovative in two ways. First, they provide a framework for comparison of past
studies on methane emissions. In a feat of graphic creativity, Brandt et al. put all of the existing studies,
bottom-up (denoted by triangles and dashes) and top-down (denoted by circles, squares, and
diamonds), on a single chart. The result helps illuminate how these two threads in the literature relate
to each other. Bottom-up studies measure facilities or components: the largest value found by any such
study was around 1Q9 g of methane emitted per year. In contrast, even the smallest of the top-down
studies, which measured the Denver-Julesberg basin, reported over 4*10' g of methane.
Brandt et al. also conduct a meta-analysis of national-scale, top-down studies of methane emissions.
The authors develop a normalization procedure to make the multitude of studies comparable. The
result indicates that the most likely range of actual methane emissions is 25-75 percent higher than the
EPA inventory indicates. This range of possible emissions is illustrated in the inset panel for Brandt et
al.'s principal graphic, which we reproduce as Figure 2. Note that for all of the studies that are national
or continental in scale, observations all lie between 1.25 and 1.75—that is 125 percent and 175 percent
of the EPA inventory.
Figure 2. Normalized comparison of top-down studies in Brandt et al.
10i4 j Ratios with common baseline (EPA GHGI)
10
13
Emissions
magnitude
{grams of 10 12
methane per
year)
10
n
10
10
0
'—
!o—
1
2 3 T 5 6
Ratio: Measured/Inventory
"T"
7
Scale of measurement
O National or continental
~ Multi-state
O Regional or air basin
-r-
8
Source: Brandt et a!.., 2014
3
-------
Figure 3 shows in red the lower and upper estimates (7- 21 Tg) of methane emissions that the
not account for in their inventory, which we are referring to as missing methane. The missing
is shown on top of the results from the EPA's latest inventory.
EPA did
methane
Figure 3. EPA methane inventory and
estimates of missing methane
Upper estimate
EPA inventory
Low estimate
Enteric fermentation
(cattle related)
¦ Natural gas systems
Landfills
Other
0
50 60 ¦ Missing methane
Tg of methane / year
Because of the limited ability of top-down studies to trace methane back to specific ground-level
sources, it is not possible to determine the origin of the missing methane with great certainty. Still,
there is reason to believe that at least some of the missing methane is coming from the natural gas
system, as there are downward structural biases in the inventory. For example, it would be reasonable
to expect that facility operators who believe they may have above-average emissions would be hesitant
to join voluntary studies. This may have a large impact on results, as there is accumulating evidence
that "super emitters" - a small number of facilities with particularly large leaks - could be a majority or a
large fraction of overall emissions. Another downward structural bias is the EPA's choice to reduce the
emissions estimated through the bottom-up procedure based on industry assertions that they have
taken voluntary actions above and beyond those required by regulations.
The large range of uncertainty remaining about the rate of emissions in the natural gas system is an
indicator of the complexity of the situation. The natural gas system is large, complex and
heterogeneous, in both engineering and geologic terms. Each natural gas-bearing basin is unique, and
there is great variation in how producers operate. Methane emissions come not only from wells
producing natural gas, but also from those mainly producing oil. Indeed, 20 percent of the nation's gas
is "associated gas" produced at oil wells. Oil wells have different emissions characteristics from wells
designed to extract primarily natural gas. The intermingling of the oil and natural gas systems also
introduces the question of how to attribute methane emissions. Some of the methane emissions from
the petroleum system should be attributed to natural gas, but determining the appropriate fraction is
challenging.
4
-------
4, Computational extensions
The Brandt et al. paper concludes that some of the missing methane is likely coming from the natural
gas system. It explores the specific possible sources of methane from the natural gas system beyond the
EPA estimates. In the supporting materials for the article, the authors develop what they call a worst-
case scenario for emissions from the natural gas system that considers the notion that all of the missing
methane is from natural gas. Under such a scenario, if 7-21 Tg of extra methane was being emitted
from the natural gas system, that would imply emission rates two to four times higher than the EPA
inventory estimate.
While concluding that some of the missing methane almost certainly originates from the natural gas
system, the Brandt et al. paper also emphasizes the continued lack of certainty regarding the extent that
natural gas emissions are underestimated. To emphasize this uncertainty, the authors consciously chose
to refrain from translating missing methane into emission rates. We also find it useful to illustrate the
potential magnitude of the problem through some further computation, including implied emission
rates for the natural gas system at different levels of missing methane.
Here, we develop four scenarios, translating the missing methane into an emission rate of methane
from the natural gas system. The emission rate is calculated by adding a portion of the missing methane
(varying by scenario) to the methane emissions assigned to the natural gas industry in the EPA's
inventory, then dividing that value by the sum of natural gas production plus total methane emission in
that scenario. We also specify the ratio of each scenario's methane emissions attributed to natural gas
systems to the corresponding value from the EPA inventory. The scenarios are shown in Table 1.
Table 1. Emission scenarios
Scenario
Implied missing methane
from natural gas systems
Ratio of scenario to EPA
natural gas system emission
Implied natural gas
system emission rate
1.
1.8 Tg
1.25
1.75%
2.
3.5 Tg
1.5
2.1%
3.
1 Tg
2
2.8%
4.
14 Tg
3
4.2%
We chose these scenarios to provide the broadest range of what seems possible in light of the work by
Brandt et al. The paper explicitly says that it is not likely that the 21 Tg of methane all comes from
natural gas, so that total amount is not considered. The upper bound analyzed is 14 Tg extra from
natural gas systems. At the low end of the range of scenarios, we analyze 1.8 Tg of extra methane
coming from the natural gas system. This would be the case if, for example, the natural gas system is
responsible for 25 percent of the lowest estimate of missing methane. Additionally, we consider two
intermediate scenarios, under which 3.5 and 7 Tg of missing methane due to natural gas systems.
5
-------
Next, we convert the methane leakage to carbon dioxide equivalent, which we use to compute an
equivalency in coal plants. Coal plants comparisons are based on annual emissions using 2012 data for a
generator of average efficiency, capacity factor and size for the U.S. fleet (a 543-megawatt generator
operating at 85 percent capacity with a heat rate of 10,444 Btu per kilowatt-hour, from the Energy
Information Administration 2013).
We use Global Warming Potential (GWP) factors to perform the conversion to C02 equivalent. GWP
factors represent the relative contribution to global warming from GHGs other than carbon dioxide,
which each have different atmospheric residence times and abilities to trap heat. All GHGs are defined
in relation to carbon dioxide, the most prevalent GHG, which is assigned a GWP of one for all time
periods.
Methane has an especially pronounced effect in the initial years and decades after it is released. Unlike
carbon dioxide, which can continue to drive warming for hundreds or thousands of years after it is
emitted, methane has an atmospheric residence time of approximately 12 years. However, while it is in
the atmosphere, methane is a very potent greenhouse gas. Moreover, atmospheric chemistry
transforms methane into carbon dioxide overtime. The most recent Intergovernmental Panel on
Climate Change (IPCC) reports GWP factors for methane of 34 over 100 years and 86 over 20 years, an
increase since the prior IPCC report that reflects improved scientific understanding.
In the past, when climate change seemed like a distant problem, using 100-year GWP values was an
accepted convention. The EPA inventory still refers to carbon dioxide equivalent without any reference
to the timeframe with the expectation that readers will assume the numbers are on a 100-year scale.
Today, with evidence of damages from climate change accumulating, there is increasing attention to
near term climate disruptions. Put differently, the value of short-term climate mitigation benefits has
been getting more attention from policy-makers. While carbon dioxide emissions will largely determine
the extent of global warming in the long run (Harvey et al.. 2013), reducing emissions of gases like
methane will reduce short-run climate damages and can be used strategically to reduce peak warming
(National Research Council 2011). Methane also contributes to the formation of ground-level ozone, so
there are local air quality benefits to emission reductions.
This issue brief presents comparisons over both shorter and longer term time periods (20-year and 100-
year GWPs). Figure 4 depicts the 20-year values in carbon dioxide equivalent (C02e) and the
comparable number of average coal plants for each of the leakage scenarios detailed in Table 1.
6
-------
Figure 4. Natural gas sector methane emissions
scenarios using 20-year GWP
2000
_>
| 1500
c
TO
g 1000
u
500
¦«§
1.4% (EPA) 1.75% 2.1% 2.8% 4.2%
Methane emissions scenarios
400
_ns
Q.
"S
o
u
The first bar represents the level of methane emissions from the natural gas sector in the EPA inventory.
An emissions rate of 1.4 percent implies emissions equivalent to 124 coal plants using 20-year GWP. A
1.8 percent emissions rate would imply emissions with a carbon dioxide equivalency equal to 31
additional coal plants beyond the basic inventory estimate, for a total of 155. Leakage of 4.2 percent
would imply additional emissions with a carbon dioxide equivalency equal to 249 more coal plants, for a
total of 373.
Figure 5. Natural gas sector methane emissions
scenarios using 100-year GWP
2000
1800
"ro
1600
3
C
1400
C
IT!
1200
01
(N
O
1000
U
H—
800
o
1-
600
2
2
400
200
0
I I
400
350
300
250
200
150
100
50
0
_ns
Q.
"S
o
u
1.4% (EPA) 1.75% 2.1% 2.8%
Methane emissions scenarios
4.2%
Figure 5 shows that, using 100-year GWP factors, the EPA estimate of methane leakage, 1.4 percent, has
a carbon dioxide equivalency equal to 53 coal plants. A leakage rate of 1.8 percent would imply
additional emissions with a carbon dioxide equivalency equal to 13 additional coal plants, for a total of
7
-------
66. Doubling EPA's leakage rate to 2.8 percent results in an additional 53 coal plants, for a total of 106.
A tripling of emissions to 4.2 percent would imply additional emissions with a carbon dioxide
equivalency equal to 107 more coal plants, for a total of 160.
5. Implications for emissions impacts of electricity from natural gas
Proponents of natural gas have pointed to the lower carbon dioxide pollution emitted from the
smokestacks of natural gas-fired electricity generators. Natural gas plants have smokestack emissions
that are roughly half those of coal-fired power plants. Yet, methane emissions from the natural gas
system significantly reduce this smokestack advantage. One of the reasons it is important to
characterize methane emissions from the natural gas system more accurately is to provide a more
accurate picture of the environmental impacts of electricity produced with natural gas. (It is worth
noting that electricity generation accounted for 39 percent of natural gas consumption in 2012.
Therefore, it is only appropriate to attribute that same fraction of the missing methane to electricity
generated from natural gas.)
Based on the new understanding of the likely range of methane leakage provided by Brant et al., it
seems very likely that substituting natural gas for coal-combustion to produce electricity actually
exacerbates climate change over the short run, i.e. 20 years, and lowers greenhouse gas emissions over
the long run, i.e. 100 years, (Alvarez et al. 2012). Being somewhat better than coal over a 100-year time
horizon is hardly a sufficient condition to conclude that natural gas can serve as the low-carbon bridge
to a clean energy future, as it is often called. In a U.S. context, it has been suggested that natural gas
use will have to peak by 2030 for the Obama administration's climate goal to be achieved (Banks and
Taraska 2013). From a global perspective, even those who extoll the virtues of natural gas have found
that if global concentrations of carbon dioxide are to remain below 450 part per million - the level that
scientists are targeting to limit the risks of dangerous climate change - then the time is very short for
natural gas to serve as a useful bridge fuel (Levi 2012).
6. Conclusion
The EPA should take steps to address clear evidence that its inventory of GHG emissions is
undercounting methane. In the short run, as part of finalizing the 2014 inventory, the agency should
make the case for a significant effort to improve the inventory of emissions from the natural gas sector.
In the longer run, the agency should develop a plan for integrating top-down data as well as new
technologies that operate at ground level that can assist in leak detection and measurement. The
federal government should be placing more emphasis in and devoting more resources to this effort.
Brandt et al.'s work illustrates the value of top-down measurements to provide evidence of overall
emission levels over large areas. The EPA should move to collect airborne measurements into its GHG
inventories. By conducting measurement campaigns, EPA will be able to obtain atmospheric data that is
more comprehensive across space and time. This will enable the agency to identify aggregate emissions
8
-------
levels with much greater accuracy and will help to improve confidence intervals. Current confidence
intervals are much too small in light of uncertainty about the true value.
Emerging technologies can link emissions back to sources, enabling the EPA to conduct an effective
ground-level measurement campaign. Infrared cameras are effective at locating leaks, and their use has
been required under a recently approved Colorado regulation. Low cost stationary detectors are also
under development. The newest detectors can locate leaks and estimate their magnitude from a
distance, which reduces the challenge of acquiring property owner permission that bedevils direct on-
site measurement.
The current oil and gas boom has been unleashed by a wave of technological innovation (directional
drilling, hydraulic fracturing, and other emerging techniques, like "acidizing"). Governments need to
keep pace with faster innovation on the regulatory side. New monitoring technologies are an
opportunity for greater accuracy, and the EPA should move quickly to use these technologies to
transform government monitoring of emissions. Better monitoring of emissions will help the EPA solve
the mystery of the missing methane and provide the best objective guidance to policymakers,
regulators, and society.
Acknowledgments
Thanks to Sonia Aggarwal, James Arnott, Adam Brandt, Eric Gimon, Hal Harvey, John Katzenberger,
Veery Maxwell, and Jeffrey Rissman for helpful comments on this paper. Any remaining errors are the
author's responsibility.
References
Alvarez, Ramon, S.W. Pacala, J.J. Winebrake, W.L. Chameides, and S.P. Hamburge. 2012. "Greater focus
needed on methane leakage from natural gas infrastructure," Proceedings of the National Academy
of Sciences 109(17): 6435-6440.
Banks, Darryl and Gwynne Taraska. 2013. U.S. Natural-Gas Use Must Peak by 2030. Center for American
Progress: Washington, DC.
Brandt, A.R., G.A. Heath, E.A. Kort, F. O'Sullivan, G. Petron, S.M. Joraan, P. Tans, J. Wilcox, A.M.
Gopstein, D. Arent, S. Wofsy, N.J. Brown, R. Bradley, G.D. Stuckey, D. Eardley, R. Harriss. 2014.
"Methane Leaks from North American Natural Gas Systems," Science 343: 733-735.
Energy Information Administration (US Department of Energy). 2013. Electric Power Annual.
Harvey, Hal, Franklin Orr, and Clara Vondrich. 2013. "A Trillion Tons," Daedalus 142(1): 8-25.
Levi, Michael. 2013. "Climate consequences of natural gas as a bridge fuel," Climatic Change 118 (3-4):
609-623.
National Research Council. 2011. Climate Stabilization Targets: Emissions, Concentrations, and Impacts
over Decades to Millenia. The National Academies Press: Washington, DC.
9
-------
Appendix C
-------
Gas Safety Incorporated (GSI)
16 Brook Lane
Southborough, Massachusetts 01772
774-922-4626 www.gassafetyusa.com
Report to
Damascus Citizens for Sustainability(DCS)
25 Main Street, Narrowsburg, New York 12764
mail to: P.O. Box 147, Milanvilie, PA 18443
Extended Report on a Preliminary Investigation of
Ground-Level Ambient Methane Levels in Manhattan,
New York City, New York
11 March 201 3
by
Bryce F. Payne Jr.1 and Robert Ackley2
[This report is subject to revision.]
EXECUTIVE SUMMARY
DCS requested that GSI extend the work effort described in our initial Report on
a Preliminary Investigation of Ground-Level Ambient Methane Levels in
Manhattan. New York City. New York (1 6 December 201 2) to assess the
practicality of developing an estimate of methane emissions in Manhattan.
Specifically the effort was to focus on providing an estimate of methane
emissions that could be used in evaluating the role of natural gas leakage in
Manhattan with respect to fossil fuel dependence, climate impacts and other
environmental and economic concerns.
Currently the greenhouse gas equivalence of methane is widely accepted as at
least 20 times the effect of carbon dioxide over a 1 00-year time frame. In
1 Consulting and research in environmental science since 1992. Associate Research Professor, Dept. Environmental
Engineering and Earth Sciences, Wilkes University, Wilkes-Barre, PA and Senior Fellow of the Wake Forest
University Center for Energy, Environment, and Sustainability, Winston-Salem, NC. bryce.payne@wilkes.edu
2 President of Gas Safety, Inc. with 30 years experience in gas leak detection and measurement, related regulatory
compliance, and training, bobackley@gassafetyusa.com
-------
-2-
other words, leakage of 1 /20th, or 5%, of the methane moving through a natural
gas production-transport-distribution system will effectively double the
greenhouse gas impact of the use of that natural gas. That is, leakage of only
5% of natural gas from point of production to point of use would eliminate any
greenhouse gas advantage of natural gas compared to other fossil fuels. More
complex efforts by others have looked into the greenhouse gas emissions
advantages of using natural gas instead of other fossil fuels. It appears that
those more elaborate efforts are settling in at <3.2% gas loss to leakage as the
maximum leakage rate at which use of natural gas retains an advantage.
Hence, the loss of even a few percent of gas during production, transport,
distribution and utilization is critically important to management and planning
of present and future national and international energy supply and utilization
systems. Therefore, it was concluded the extended GSI work effort should be
focused on the need to assess total methane emissions. The available data was
from Manhattan. Among the production, transport, local distribution and
utilization systems, this work addressed the collective effect of only local gas
distribution and utilization systems, along with any other methane sources that
might be present in Manhattan .
GSI efforts for this extended report focused on three objectives: (1) find
existing estimates from industry, government or other sources, of the amount
of methane being released in Manhattan, (2) develop such an estimate from the
ground-level methane data collected during our preliminary investigation of
methane levels in Manhattan, and (3) compare those estimates and consider
their implications with regard to broader environmental and economic
concerns. Since this investigation was limited to Manhattan (augmented with
comparative data from the Bronx, and other areas across New York and
Connecticut), ConEd is the relevant gas distribution company.
An examination of existing estimates, or methods for estimating, methane
emissions led to the conclusion that such estimates have little basis in actual
data. Natural gas companies are required to file yearly reports of Lost- and-
Unaccounted-for (LAUF) gas. Presumably these reports would approximate the
amount of gas leaked from the pipelines and other infrastructure of the
reporting companies. However, the meters in those gas systems are only
required to be accurate to ±2%. Each such system may contain hundreds of
thousands of meters. Each meter is subject to normal wear and tear. Another
problematic issue is the reported LAUF gas volume may incorporate other gas
volumes by rule, contract, regulation, or for other administrative reasons.
Consequently, the annual reported LAUF gas volumes should not be regarded
as reliable estimates of the amounts of gas actually lost or emitted to the
atmosphere. However, since the LAUF gas volume is ultimately based mostly
on measurements using meters that are accurate to ±2%, it follows that long-
-------
-3-
term average LAUF values should provide a reasonably meaningful mean with a
±2% variability. A ten-year average LAUF for ConEd was 2.2% with a range of
0.4 to 4.3%, i.e, ±2% variability. The 1 O-year-average-LAUF based estimate of
annual methane emissions for the entire ConEd system was 2.2% or about 6.6
billion cubic feet per year.
The apparently most widely used method for estimating gas leakage and
methane emissions from gas pipelines appears to be from a 1 996 report by the
U.S. Environmental Protection Agency and the Gas Research Institute (EPA/GRI).
Estimates generated using the EPA/GRI 1996 method have such a wide
confidence interval (±65%) that their general accuracy and usefulness is
questionable. The report recognizes the likely importance of gas leaks that are
undetectable by the standard industry leak detection practice, but the
estimation method makes no attempt to account for such undetectable leaks.
Finally, a related report of a more thorough study of cast iron pipelines in
Brazil, suggested that the EPA/GRI method may provide estimates that are too
low by almost half. Application of the EPA/GRI method to the pipeline statistics
for the entire ConEd system generated an estimated methane emissions rate of
1 billion cubic feet per year, which can be meaningfully compared to the 10-
year average ConEd LAUF gas estimate of 6.6 billion cubic feet per year. Since
most leakage in gas delivery systems occurs from the pipes in the system, such
a disparity between the EPA 1 996-based estimate for ConEd pipeline leakage
and the 1 0-year average ConEd LAUF gas volume would seem to indicate
problems in one or both of those estimates.
During the research for this Report, we thoroughly reviewed the methane data
collected by GSI during the previously reported Preliminary Investigation of
Ground-Level Ambient Methane Levels in Manhattan. We also reviewed the
meteorological literature and meteorological data available for Manhattan.
Based on that information we developed a simple model (patent pending) that
could process our preliminary Manhattan methane data and meteorological data
from local sources to generate a preliminary estimate of total methane
emissions in Manhattan. The resulting estimate was the flow of methane to the
atmosphere from all sources in Manhattan. Such an estimate can be used to
assess the relative importance of those emissions in terms of methane as a
greenhouse gas (GHG) and the relative impact of gas service/use in Manhattan
in a broader climate/GHG context. Wherever reasonable in the application of
the model, input values were selected conservatively, so that any errors in the
result should be to the low side.
The resulting methane emissions estimate for Manhattan alone was 8.6 billion
cubic feet per year, or about 2.86% of the BOO billion cubic feet of gas handled
by the entire ConEd system each year, even though Manhattan comprises only
-------
-4-
about 5% of the land area and one-third of the customers in the ConEd service
territory. There are also substantial losses that occur in the natural gas system
before natural gas reaches the ConEd distribution system. It, therefore,
appears inevitable that the loss of gas in the system serving NYC via ConEd is
above the simple critical level of 5%, and well above the more elaborately
derived critical levels of <3.2%. That is, the methane leakage in the system
serving NYC through ConEd is likely already at a level where the methane
leaked has as much or more climate impact as the remaining approximately
95% of the gas that is actually usefully burned by consumers in NYC. This
necessarily raises doubts about the claimed value of natural gas as a "clean
bridge fuel". Further work should be done to verify the findings we report here
and to identify specific methane sources, as well as to improve natural gas leak
prevention and management. Furthermore, the evidence suggests that leakage
from natural gas systems has a more substantial role in climate change than
was believed that has only recently begun to be appreciated.
Panoramic image looking south from the upper deck of the Top of the Rock' observation deck
on Rockefeller Center. Image taken and assembled by Daniel Schwen on Dec 6th, 2004.
{GFDL Wikipedia}
INTRODUCTION
In our initial report (dated 16 December 201 2) on the preliminary investigation
of ground-level ambient methane levels in Manhattan, New York City, New York
we stated, "Further work is needed to determine whether an approximate estimate
of the amount of methane being released to the atmosphere can be developed
from the data generated by this preliminary methane survey." To that end our
efforts have focused on three objectives: (1) find existing estimates of the
amount of methane being released in Manhattan from industry, government or
other sources, (2) develop such an estimate from the ground-level methane
data collected during our preliminary investigation, and (3) to compare those
estimates and consider their implications with regard to broader environmental
and economic concerns. Since this investigation was limited to Manhattan
(augmented with comparative data from the Bronx, and other areas across New
York and Connecticut), ConEd is the relevant gas distribution company.
-------
-5-
Available Estimates of Methane Emissions in Manhattan
There are readily available documents that imply measurement-based estimates
of methane (natural gas) releases in Manhattan have been developed.34
However, review of those estimates leads to the conclusion that they are all
largely based on other estimates, some periodically updated, but apparently
never actual measurements of gas emissions in the field. This is presumably
due in part to the historical lack of readily available, reliable approaches to
actually measure methane concentrations and calculate methane emissions
under field conditions.
LAUF Gas
Among the more prominent of such estimates-based-on-other-estimates
would seem to be the LostAnd Unaccounted For(LAUF) gasthatcompanies are
required to report to the New York State Department of Public Service (NYSDPS).
Actually, thereported LAUFisacalculated numberthat includes volumes
actually measured by meters in the gas distribution system along with various
add-ins and deductions that are matters of contract, regulation, or used for
operational accounting reasons. In addition to the arbitrariness of the add-ins
and deductions, gas meters are only required to be accurate to ±2%.
Malfunctions leading to metering errors of more than 2%can be expected to
occur. It is importantto realizethattheestimation and reporting of LAUF gas
was never intended to represent actual losses of gas from the gas distribution
system, but to facilitate annual reconciliation of costs for gas purchased to
revenues for gas sold while providing incentive to minimize actual lossofgas.5
The reliability of LAUF numbers as estimates of actual gas losses is easily
appreciated in the following statement (with original footnotes) found in a New
3 ConEdison Gas Long Range Plan 2010-20B0, December 2010 [accessed at http://
www.coned.com/Publiclssues/PDF/GLRPl 21 Oc.pdf], and various ConEd annual and other
reports.
4 Inventory of U.S. Greenhouse Gas Emissions and Sinks 1 990 - 2009, USEPA, April, 201 1.,
Annex 3 (PDF) (232 pp, 9.6 MB) - Methodological Descriptions for Additional Source orSink
Categories, [http://www.epa.gov/climatechange/ghgemissions/usinventoryreport.html].
5 NYS DEPARTMENT OF PUBLIC SERVICE, STAFF WHITE PAPER ON LOST AND UNACCOUNTED FOR
(LAUF) GAS, issued January 27, 201 2. [White paper accessed at http://www.google.com/
search?client=safari&rls=en&q=NYS+DEPARTMENT+OF+PUBLIC+SERVICE,+STAFF+WHITE
+PAPER+ON+LOST+AND+UNACCOUNTED+FOR+(LAUF)+GAS,+Hearing+Exhibit+No.
+76,+GRP-l 5&ie=UTF-8&oe=UTF-8]
-------
-6-
York State Department of Public Service Staff White Paper on Lost and
Unaccounted for (LAUF) Gas3 (NYSEG is New York State Electric and Gas
Corporation):
"Negative Losses
Staff must address negative losses because NYSEG12 has experienced consistent
negative losses for the past 3 years. Negative losses are physically impossible.
However, consistent year to year calculated negative losses are possible when the
offset13 between the set of meters reading gas in and the set of meters reading gas
out is negative and the natural variability is less than that offset. Additionally,
natural variability in the LAUF can produce negative losses in some years for LDCs
whose offset is positive.
12 Case 09-G-0669
13 Two sets of meters will never provide the same measurement. The difference between
those two measurements is defined as offset."
Note: LDCs are Local Distribution Companies
NYSEG LAUF gas values over those "past three years" (2008-2010) averaged
-0.359%, while the ConEd average LAUF for the same three years was +1.249%.
NYSEG is not ConEd, but gas metering and related LAUF errors inevitably affect
the reported LAUF gas amounts of every company and probably in different and
unforeseeable ways that change from year to year. Unaccounted for gas
estimates are also reported annually to PHMSA6. When ten years (2002-201 1)
of those reported values were examined for this report, they were not the same
as those stated in the NYS DPS Staff White Paper3, presumably due to different
reporting requirements. Though consistently low, the NYSEG unaccounted for
gas reported to PHMSA, were never negative, ranging from 0.1% to 0.3% for the
eight years 2004-201 1. Though not implausible, such consistent and low
numbers are interesting given that meters used in gas systems are only
required to be accurate to ±2%. For the ten years 2002-201 1, ConEd reported
annual unaccounted for gas percentages ranging from 0.4-4.3. In contrast to
the consistently low numbers of NYSEG, the ConEd numbers appear to have a
variation of very close to ±2% around a mean of 2.2%. Coincidentally, 2.2% also
happens to be the mean of all unaccounted for gas percentages reported to
PHMSA from 2002-201 1, though among those numbers individual annual
reports ranged from -28% to +109%. Such examples serve to illustrate that
LAUF numbers provide little if any useful insight into the actual amounts of gas
lost from companies' gas distribution systems at any given time, or over a given
year. Still, it is helpful to consider a bit further the implications of the average
6 PHMSA - US Department ofTransportation Pipeline and Hazardous Materials Safety
Administration. Lost and Unaccounted for Gas reports accessed at http://
www.phmsa.dot.gov/portal/site/PHMSA/menuitem.ebdc7a8a7eB9f2e55cf20Bl 050248a0c/?
vgnextoid=a872dfal 22al dl 1 OVgnVCMl 000009ed07898RCRD&vgnextchannel=3430fb649a2d
cl 1 OVgnVCMl 000009ed07898RCRD&vgnextfmt=print
-------
-7-
unaccounted for gas percentage of 2.2%.
A Little Bit Matters
A loss of 2.2% might seem almost trivial. Each gas consumer, based on the
required accuracy of the meter that measures gas consumption, can expect that
they may be over or undercharged by as much as 2% anyway. Why, then,
should anyone concern themselves with a loss of a few percent over the
distribution system as a whole? A first answer would be a fair allocation of the
monetary cost of the lost gas. In 2011 ConEd had total gas sales and
transportation revenues of around 1.5 billion dollars, 2.2% of which amounts to
33 million dollars. That is a substantial amount of money and has to be
accounted for and fairly allocated, a process that is regulated by the NYS
Department of Public Services. Again, though, in the grand scheme of things,
the consequences for each customer are relatively minor, only 0.2% more than
the ±2% of metering accuracy. So, we are still left with the question, why does
such a seemingly small amount matter?
There are two closely related reasons. One, it remains that, regardless of the
reporting of the amounts of lost and unaccounted for gas, those reported
amounts do not seem to provide a reliable indication of the actual losses of gas
that are occurring. Two, when methane, which makes up over 90% of natural
gas, escapes from the distribution system it can accumulate to pose direct risks
of injury and property damage. A less obvious but greater global concern is the
role of methane as a potent greenhouse gas. Any leakage of methane poses an
effectively invisible, but potentially substantial threat to human health and the
environment. These reasons provide a means of understanding why the actual
amounts, and locations, of even seemingly small gas losses matter.
Even small natural gas leaks in confined spaces are dangerous, posing
explosion and asphyxiation hazards. When a small underground gas leak finds
a pathway to an enclosed space, such as a manhole, the gas can accumulate to
explosive levels (5%-l 5% methane). Basements and other poorly ventilated
spaces can also accumulate leaked gas to hazardous levels. Explosions related
to such accumulations of leaked gas, though not common, are recurrent
wherever natural gas is used. In addition, where even relatively small amounts
of gas are leaked into the soil for extended periods, vegetation will be
damaged, loss of urban trees being a common impact. Still, the ConEd record
of gas safety with regard to direct hazards is relatively good.
ConEd, like other gas companies, has a routine program to detect, manage and
repair leaks. However, the objective of such leak control programs is to detect
-------
-8-
leaks, not measure the amount of gas lost through them. Such measurements
would be impractical, especially for the potentially very large numbers of very
small leaks that can be expected to develop in pipe systems that contain
substantial amounts of old pipe. Over 70% of the cast iron pipe in the ConEd
system is over 1 00 years old, and almost all was installed before 1 930, i.e., is
more than 80 years old.3
EPA Leakage Estimates for Natural Gas Pipelines
In this scenario, we are left with potentially large numbers of small leaks, and
smaller numbers of larger leaks in gas pipe systems. Measurement of the gas
losses that occur through such leaks is in practical terms impossible. Most of
the small leaks will never be identified, let alone measured. How, then, does
anyone arrive at some reasonable estimate of how much gas is being lost? In
1 996 the U.S. Environmental Protection Agency (EPA) released an approach for
estimating such losses.7 This approach is of considerable importance because
it has become the basis for international estimates of methane/natural gas
leakage as well.8
The EPA approach7 is relatively simple, based on 4 types of pipe materials, cast
iron, unprotected steel, protected steel, and plastic. The estimated leak rates
for the 4 types of pipe were based on data collected in a 1 992 study by the EPA
and the Gas Research Institute (GRI). The length of pipe of a given type in a
system is multiplied by an estimated leak rate for a given length of that type
pipe. For cast iron pipes, the oldest and leakiest type, the estimated leak rate is
in standard cubic feet per mile of pipe per year (scf/mile-yr). That study
looked at a total of 21 samples of cast iron pipe. The estimated methane leak
rate for cast iron pipe was 399,867 scf/mile-yr (with a 90% confidence interval
of 227,256). This was reduced by another factor intended to account for the
amount of methane that would be biologically oxidized in soil before escaping
into the atmosphere to produce a "Methane Emission Factor" for each type of
pipe. After that reduction the estimated emission factor for cast iron pipe
became 238,736 scf/mile-year (with a 90% confidence interval of 1 52,059).
The 90% confidence intervals and numbers of samples are mentioned in this
discussion because it is important to understand how imprecise these estimates
7 EPA/GRI. Methane Emissions from the Natural Gas Industry. Volume 9: Underground
Pipelines. June 1996. http://www.epa.gov/qasstar/documents/emissions_report/
9 underqround.pdf.
8 IPCC. 2006 IPCC Guidelines for National Greenhouse Gas Inventories, http://www.ipcc-
nqq i p. iqes.or. ip/public/2006q l/index. htm l>.
-------
-9-
are. The numbers seem so imprecise that their usefulness seems questionable.
The statistically strongest data set in EPA/GRI7 was that for cast iron pipe. The
data indicates that there is only 90% confidence that the true mean leak rate for
cast iron pipe is somewhere in the range of 399,867±65%, that is, somewhere
between 1 72,000 and 626,000 scf per mile of pipe per year.9 The 90%
confidence level seems low for an estimate that has implications as broad and
important as this one. Accuracy is critical in estimating emissions of the
second most important greenhouse gas, methane, when these estimates are
being used in both national and international estimates for climate change
modeling and planning of mitigation and response measures..8 At least a 95%
confidence interval would seem more traditional and appropriate to the
purpose. However, back calculation from the 90% confidence levels and sample
numbers in EPA/GRI7 report indicate that the 95% confidence intervals would
extend below zero for unprotected steel and plastic pipes, and would approach
zero for protected steel. In fact, in the case of plastic pipe, with a high
variability (range 0.008 to 61 std.cu.ft. per leak per hour) and the lowest
number of samples (N=6), even at the liberal 90% confidence level, the lower
limit of the confidence interval was -60,000 std.cu.ft. per leak per year,
implying the impossible situation that relatively large amounts of gas could be
taken in instead of emitted by leaks in plastic gas lines. One might reasonably
set aside the issue of implied negative leak rates, and allow that leak rates
below zero cannot occur. Even from this perspective, one is left with the
predicament that the EPA/GRI7 data for plastic pipe do not distinguish at a 90%
confidence level between 260,000 scf per leak per year and no leak at all.
A Leakage Estimate from Comgas in Brazil
The EPA estimate approach is still the international norm, but more recent work
reported out of Brazil provides a different picture.10 That study by the Brazilian
natural gas distribution company Comgas used a different approach to
selecting samples, and a very conservative approach to disregard all
suspiciously or inexplicably high leak rates. The Comgas study was apparently
continuous from 2005 through at least 2009 as part of a pipe system upgrade
program. Consequently, pipe sections selected for testing were each almost
9 EPA/GRI7 is not clear regarding whether a one-sided or two-sided confidence interval was
used. The statement, "an overall accuracy of ±65% based on a 90% level of confidence"
suggests a two-sided confidence interval was used, but repeatedly in footnotes to tables "upper
bound minus the mean" may indicate a one-sided confidence interval was used. We assumed
that all confidence intervals referred to in EPA/GRI7 were two-sided.
10 Carey Bylin et al. 2009. New measurement data has implications for quantifying natural gas
losses from cast iron distribution mains. Pipeline and Gas Journal, (www.pgjonline.com).
-------
-10-
certainly considerably larger than the minimum 20-foot sections in the EPA/GRI
1 992 study11 and were effectively more randomly selected. Random selection
based on work scheduling without regard to prior detection of leaks combined
with measurements of longer pipeline segments means the Comgas study
would more likely measure total leakage, where the EPA/GRI approach was
based on detection of leakage before testing. In the course of the Comgas
work in Brazil, 91 2 pipe sections were tested, compared to only 21 in the EPA/
GRI 1 992 study. The Brazilian cast iron pipe system was reported to be
otherwise comparable to the U.S. cast iron system studied by EPA/GRI in 1 992.
The Brazilian cast iron pipe, however, would likely be considerably younger
than that in the ConEd system in which 70% of the cast iron pipe is over 100
years old. Instead of a methane leak rate of 399,867 scf/mile-yr the Brazilian
study found a leak rate of 750,51 3 scf/mile-yr. It is interesting that though the
Brazilian study may be regarded as contrasting with the EPA/GRI, in fact, it
actually is statistically compatible. We back calculated the standard deviation of
the EPA/GRI7 cast iron pipe results and concluded the 750,000 scf/mile-yr
appears to be within 99% confidence bound of the EPA/GRI7 study. That is, the
findings of the two studies do not seem to conflict. The Brazilian is simply a
more robust, larger study that should provide a more accurate estimate and is
statistically compatible with the EPA/GRI estimate.
Yet, even the higher Brazilian numbers may be too low because data from pipe
sections with suspiciously or inexplicably high leak rates (>1,991,444 scf per
mile per year) were excluded. The excluded data was 1 5.4% of the total data.
The concern behind that elimination of high leak data was that such data could
be caused by measurement procedural problems in the field or unmapped
service lines connected to the cast iron mains. It would seem likely that leaks of
this size would result in noticeable mercaptan odors and consequent leak
reports. Nevertheless, it also seems reasonable that such large leaks may
develop slowly and exist for some time before odor motivates reports of
suspected leaks, though 1 5.4% of pipeline test sections seems implausibly high.
The concern that such high data are due to procedural difficulties or unmapped
services seems reasonable, but one avoided at the risk of entirely missing some
actual large leaks. For example, if the tested sections are relatively long, there
could be several moderate sized leaks that collectively cause leak rates above
11 The actual lengths of cast iron pipe sections were apparently variable and not clearly
specified in the 1 996 EPA/GRI7 report of the results of the 1 992 EPA/GRI study of pipe leak
rates: (on page 20 of that report) "The segment to be tested was either: 1) a service which was
isolated ... at the service-to-main connection and the customer's meter, 2) a short segment of
main (at least 20 feet long) containing the detectable leak which was isolated by capping both
ends, or B) a long segment of main containing multiple leaks...isolated by capping off each end.
... For cast iron pipes, a segment test approach was used since many undetected leaks are
known to exist in cast iron."
-------
-11 -
the Comgas sample rejection level. Without knowing the lengths of the Comgas
test sections, it is not possible to resolve this doubt. For present purposes, it is
sufficient to let the Comgas test results stand as reported.
Estimates of Methane Leakage for ConEd based on EPA/GRI and Comgas
Reports
Most of the oldest and leakiest pipe in many natural gas systems is cast iron.
About 30% of the mains in the ConEd pipe system are cast iron, with another
30% unprotected steel, the next leakiest type. Now, using the EPA Methane
Emission Factor extrapolation approach would seem reasonable enough, in fact,
a practical necessity given the amount of underground pipe in natural gas
distribution systems. For example, ConEd has about 1 300 miles of cast iron
mains, with similar amounts of unprotected steel, all of which feed eventually
into hundreds of thousands of smaller service lines. Clearly the amount of gas
leaking from each segment of such an extensive gas pipe system cannot be
monitored continuously.12 Given the soil conditions under the streets of
Manhattan, biological oxidation of methane is probably limited. So, if one
applies the (no soil methane oxidation) EPA Methane Leakage Factor of
(rounded) 400,000 scf/mile-yr for cast iron mains to the 1 300 miles of cast
iron pipe in the ConEd system one arrives at estimated methane emissions of
520,000,000 scf/yr. If one uses the Brazilian Comgas cast iron pipe leak rate
this becomes 975,000,000 scf/yr, which could also be too low.
Other Leak Sources and Other Estimates
One could similarly generate estimates for the other likely sources of gas
leakage in the ConEd system in accordance with EPA estimating methods. In
fact, beginning in 201 0 ConEd, along with most other large emitters of
greenhouse gases, has to file a report of estimated emissions of GHGs,
including methane, with the EPA every year. However, during the preparation of
this report only the 2010 GHG emissions report for ConEd had been filed and
released by EPA. That 201 0 ConEd report contained only volumes of natural
12 In fact, in general any given section of pipe is checked every 1 -3 years. Type 3 leaks that are
detected but do not present an explosion hazard at the time of detection, and are deemed not
likely to subsequently present such a hazard, are not repaired but put on a somewhat more
frequent inspection schedule to assure they do not increase to a hazardous level. That is, they
are left to continue leaking until they increase to an explosion hazard level or are repaired
under routine leak repair efforts. Such unrepaired Type 3 leaks effectively release methane
emissions without a control effort because they do not present an immediate or foreseeable
explosion hazard.
-------
-1 2-
gas delivered, which totaled 286,962,094,000 scf. The number of potential
sources of leaked methane, besides cast iron pipe, in the ConEd system is
large, perhaps explaining why the 2010 ConEd GHG emissions report to EPA is
empty. For the purposes of this report, a simpler approach may serve the
immediate purpose of showing that presently reported numbers are not reliable
and approaches to actual measurement are needed.
Consider in this regard that through the EPA Natural Gas STAR program ConEd
has been credited with reducing methane emissions by 4,393,61 3,000 scf
cumulatively since 1 993. That 1 8-year (or so) cumulative reduction barely
makes up for somewhere between 4 and 8 years of the estimated ongoing
leakage from cast iron pipes alone, depending on the leak rate factor used.
ConEd reported to the EPA GasSTAR program that in its best single year, 2008,
it reduced methane emissions by 1 58,795,000 scf. That is, in its best year,
ConEd eliminated the equivalent of barely 30% of just one year of losses from
the cast iron pipe alone. So, given there are still 1 300 miles of cast iron pipe in
the ConEd system, and there are many other potential leaks in the ConEd
system, ConEd may well be losing ground with respect to overall net methane
emissions. Further, if one considers that the total gas handled annually by
ConEd amounts to about 300,000,000,000 scf1, then the estimated cast iron
pipe leakage alone amounts to in the range of 0.1 7-0.33%, and this estimate
could still be low.
When Is a Leak a Leak?
When It Is Detectable.
Another matter worth considering is the functional definition of a leak. In the
ConEd Long Range Gas Plan (201 0)1 there is the following statement (including
associated original footnotes).
"Con Edison also performs extensive leak repairs annually and has managed to
reduce the backlog of leaks .... In 1 988, the gas leak backlog was just over 1 5,000
leaks and year-end 2009 leaks were under 1,400. Most of the leaks in the leak
backlog are Type 323 leaks which are not hazardous. We enter each winter with less
than 1 00 hazardous leaks. Gas leak repairs are a major commitment of our O&M
expenses. Con Edison has the highest amount of leak reports issued annually of all
NYS utilities. Con Edison has committed to the NYS Public Service Commission that
ConEd will maintain a leak backlog of less than 1,60024 leaks at the end of the year.
23 A Type 3 leak is not immediately hazardous at the time of detection and can be
reasonably expected to remain that way. However, Type 3 leaks shall be reevaluated during
the next required leakage survey or annually whichever is less.
24 NYS PSC mandates a leak backlog less than 1 600 leaks at the end of the year."
-------
-1 3-
The contention of ConEd regarding the total number of leaks may be
reasonable given industry leak detection practices, but not at all accurate in
terms of actual total pipe leakage. A similar statement has to be made with
respect to the previously discussed 1 996 EPA/GRI report7 providing the now
widely used methane emission factors for gas pipelines.
Cast iron gas distribution (pipe) mains have been in the ground longest among
all the predominant pipe types in the commercial natural gas system. EPA/GRI7
reported that cast iron pipelines were found to be much leakier than the
pipelines of the other pipe materials. The high leakage from cast iron pipes is
due to large number of small leaks, "For cast iron pipes, a segment test
approach was used since many undetected leaks are known to exist in cast
iron." EPA/GRI7 also reported experiments indicated 40.3% of the methane
leaked from cast iron pipes was oxidized during its rise to the soil surface, but
only 1.8-3.0% for the other pipe types. Soil methane oxidation rates measured
around cast iron pipes were much higher than for other types because the
methane leakage is spread more widely around and along cast iron pipes. For
the other pipe types, detected leaks tended to be larger but fewer in number
resulting in more concentrated methane and less oxidation in the soil.
So, when, then, is a leak a leak? When gas escapes from a pipeline is it like the
proverbial tree falling in the forest? When gas escapes from a pipeline is it a
leak, or is it not a leak until the gas company detects it? The following quote
from the EPA/GRI report7 explains the typical industry approach to detecting
gas leaks.
"Gas distribution operators use leak detection procedures to locate and
classify leaks for repair. To identify a leak in a section of pipe, a portable
hydrocarbon analyzer or flame ionization detector (FID) was used to screen
immediately above the ground level while walking the pipeline. Any
excursions above the background level (typically 2-3 ppm) may indicate a
nearby leak."
However, the EPA/GRI7 report also states that "many undetected leaks are
known to exist" in cast iron gas mains. That is, there are undetectable leaks,
and potentially a lot of them. Again quoting the EPA/GRI7 report (page 20),
"This technique was based on testing leaks which are detected using leak
survey procedures (i.e., detected leaks), and may exclude smaller or more
diffuse leaks that are not detected at the soil surface."
Now, having established there are undetectable leaks, and since undetectable
-------
-14-
leaks are undetectable, they are not included in the leak counts of ConEd, or
any other gas company using a similar leak detection method. Similarly, since
this method was used in the EPA/GRI7 pipeline leakage study to select pipe
sections for leak testing, whether or not it accounts for any undetected leaks is
unclear. That report states,
"The leak flow rate measurement used should have accounted for all leaks in
a pipe segment. ... The segment of pipe tested was also surveyed to
determine the number of detected leaks and the corresponding
concentration of methane detected for each leak in the segment."
However, it is not clear whether or how this survey "to determine the number of
detected leaks" might have included "undetectable leaks".
So, we are left with data in industry records and the widely used EPA/GRI7 study
results that by default do not seem to address "undetectable" leaks even
though those records and that report clearly indicate substantial amounts of
such leaks do occur. At least we do know that a leak is a leak no matter how
small.
A Consideration of Undetectable Leaks
In Cast Iron Pipe
At this point one may wonder what then might an undetectable leak be like and
what difference, if any, might such leaks make? The question would seem to
resolve to how many undetectable leaks might there be that would escape
detection by the typical industry leak detection method. Leaks are usually
detected by surveying at the ground surface above a pipe with an FID
instrument set to alarm if methane (actually combustible gas) levels rise above
background levels. EPA/GRI7 accepted and included in their emission factors an
estimate by Southern Cross Corporation that 1 5% of detectable leaks are simply
missed using the standard leak survey. It would seem to make sense that those
1 5% might be predominantly smaller, hence, harder to detect leaks.
Actual individual leak data were not provided in the EPA/GRI7 report except for
the 6 data points for plastic pipe. The lowest leak measured, hence,
presumably detected, was 0.008 scf per leak per hour. It is not clear, however,
that this was a leak that actually allowed detection as the next nearest leak rate,
0.700 scf per leak hour, was approaching 100 times larger. EPA/GRI7 reported
that this 0.008 scf per hour leak value was a potential statistical outlier.
Coincidentally, it also happens to be the smallest of 6 data points, and,
-------
-1 5-
therefore, comprises roughly the bottom 15% of the leaks, i.e., the percentage
estimated to be routinely missed in leak surveys. So, if the 0.008 scf/leak-hour
value is disregarded, among the remaining five data points, the next highest 3
fall in the range of 0.7-1.62 (average 1.15) scf/leak-hour. Since these are the
only data immediately available, we will assume for this discussion that the
smallest leak that can be reliably detected using the industry leak detection
method will have a leak rate of 1 scf/leak-hour.13 As discussed below, it
matters little whether the actual undetectable leak is 1 scf per hour or
considerably lower.
It would seem to follow that if two 1-scf-per-hour leaks were next to each
other, then at the soil surface they would present the same methane
concentration as one 2-scf-per-hour leak. That is, they would be detectable.
So, then, at what distance of separation would they cease to be detectable? Gas
Safety, Inc. experience with gas leak detection indicates that under a paved
surface small leaks are detectable over a surrounding, roughly circular area in
the range of 20-25 feet in diameter, and about half that if the soil surface is
not paved over. Recall the test sections in the EPA/GRI7 study were around 20
feet which would, therefore, imply that small (<1 scf-per-hour) leaks
separated by more than 20 feet would not likely have been detected or
measured in that study. To provide some notion of what such leaks might
mean, one could assume there ought to be a range of such small undetectable
leaks that should vary from just more than zero to just less than 1 scf per hour,
which would generate an average undetectable leak size of 0.5 scf per hour.
Because undetectable leaks are undetectable, there is at present no data that
provide direct indications how many there might be per length of pipe,
regardless of the material the pipe is made of.14 Nevertheless, a rough
indication can be extracted from the data in the EPA/GRI7 report. For ten
reporting gas distribution companies, there was an average of 1.38 leak repairs
per mile of cast iron pipe. It follows that if a repair were undertaken, then it
was because a detectable leak had been found. This is actually a conservative
approach because a repair implies a detected leak, but not all detected leaks
are repaired (within a year of detection). EPA/GRI7 estimated the average
13 Based on decades of experience in gas pipeline leak detection, Gas Safety, Inc, considers
such small leaks unlikely to be detectable by conventional gas leak surveys in open field,
unpaved soil surface conditions. In urban settings, i.e., where gas lines are under paved
surfaces that can cause methane to accumulate in the soil or in underground channels or
spaces, a larger proportion of such leaks might be detected. The urban/rural setting of the
EPA/GRI7 sampling sites was not specified.
14 Except for the Comgas study7 in Brazil regarding leaks from cast iron pipes, implications of
which are discussed later in this report.
-------
-16-
number of active detectable leaks per repair was 2.14. Converting 1.38 repairs
per mile to the distance between detected leaks (repairs) yields one detected
leak for every 0.725 miles of pipe. Applying the EPA/GRI7 estimate of 2.1 4
actual detectable leaks per repair reduces the distance between detectable
leaks to 0.725/2.1 4=0.339 miles. Since the (presumably) total leak rate for
cast iron pipe was 399,867 scf per mile per year, the total leak rate for the
average length of pipe between two adjacent detectable leaks, i.e., 0.339 mile,
would be (0.339 X 399,867) = 1 35,469 scf per year.
We are trying to develop some understanding of the potential importance of
undetectable leaks. The EPA/GRI7 cast iron leakage rate supposedly includes
both detectable and undetectable leaks. So, if we deduct the rate for detectable
leaks in cast iron pipe from the total leakage, we should have the rate for
undetectable leaks. Unfortunately, there was no reported leak rate per leak in
cast iron pipe because, as previously discussed, cast iron pipe typically has a
large number of small leaks. As an alternative, we used the leak rate of 52,748
scf per leak per year for the most similar pipe, unprotected steel. Each
detectable leak is on average 0.339 miles from the next, and each 0.339 miles
of pipe has a total leakage of 1 35,469 scf per year. The undetectable leakage
should be the difference between the total leakage (1 35,469 scf/yr) and
leakage from the detectable leak (52,748 scf/yr), which is 82,721 scf per year.
This then is an estimated average leakage from undetectable leaks for the pipe
between each two detectable leaks, which occur on average every 0.339 miles.
Converting this undetectable leakage rate to leakage per mile per year yields
244,000 scf per mile per year.
This volume of leakage would be accounted for by one undetectable 0.5-scf-
per-hour leak every 95 feet along the cast iron pipeline. Perhaps, though, the
actual undetectable leak size is smaller. Even if the average undetectable leak
were smaller, say, 0.2 scf per hour, then the interval between undetectable
leaks that would account for 82,271 scf/0.339 mile per year would be 39 feet,
still farther apart than the likely 20-foot interval that might make 1-scf-per-
hour leaks detectable and well beyond the ends of the 20-foot test segments
used in the EPA/GRI7 study. So, it matters little whether the threshold for leak
detection is 1, 0.5 or 0.2 scf/hour, the implications of undetectable leaks
remain large, at least for cast iron pipe. With regard to the plausibility of this
estimate of leakage from undetectable leaks in cast iron pipe, one may consider
that adding this 244,000 scf per mile per year to the EPA/GRI7 estimated
400,000 scf per mile per year (presumably based on pipe sections with
detectable leaks) generates a total estimated leakage of 644,000 scf per mile
per year, still well below the 750,000 scf per mile per year total leakage actually
measured in the Comgas study in Brazil.
-------
-1 7-
Undetectable Leaks In Pipelines Made of Other Materials
This potential importance of undetectable leaks cannot be simply ruled
inapplicable to pipes made of other materials. There seems no reason to rule
out occasional minor manufacturing defects, damage during installation and
due to natural underground processes and animal and human activities after
installation. Indeed, unprotected steel is subject to corrosion problems, as is
protected steel, though to a lesser degree. The question becomes, then, how
to generate an estimate of the potential importance of undetectable leaks in
steel and plastic gas lines. One approach would seem to be to again exploit the
logical association of repairs to detected leaks. It was estimated above that
leaks as large as 1 scf/hour and as close together as every 20-25 feet would
likely be undetectable using the typical industry leak detection method. Once
again referring to EPA/GRI7, the reported repair interval for unprotected steel
pipeline was 1.09 repairs per mile per year, and 0.08 for both protected steel
and plastic. These can be converted, as above, to miles between adjacent
repairs, which are 0.91 7 miles for unprotected steel and 1 2.5 miles for both
protected steel and plastic. Now, it would seem reasonable to conclude if pipe
injury/defects/etc. were causing detectable leaks in cast iron, then
undetectable leaks in other pipe materials will ultimately be due to the same
causes. So, if leaks have the same causes in all pipe materials, then the ratio of
detectable leaks to undetectable leaks should be reasonably similar for all pipe
materials.
Applying this same-ultimate-causes-for-leaks reasoning and extrapolating the
estimated undetectable leakage rateforcastiron pipelinesto unprotected steel
pipelines yields an effective distance between detectable leaks of 0.429 miles,
and an estimated leakage from undetectable leaks of47,543 scf per year for
each 0.429 miles of pipe, or 1 1 1,000 scf peryear per mile of unprotected steel
pipeline. Extrapolatingtheaboveapproach indicatesflowsfrom undetectable
leaks are likely to be
-------
-1 8-
to EPA based on factors given in EPA/GRI7, the reality is we have little reason to
believe any of these estimates provide a reliable indication of how much natural
gas is leaking from natural gas distribution systems, or of how much methane
that leakage is releasing to the atmosphere. Hopefully it is at this point
obvious to the reader that actual identification and measurement of every gas
leak, or even leakage of gas from every segment of gas pipeline in service, is an
impossible, and perhaps meaningless task. In the end there remain three
objectives:
1. Fair and reasonable allocation of unaccounted for costs in the natural gas
public service system.
2. Prevention of hazardous situations related to accumulation of leaked gas
to levels that are explosive or asphyxiating (to humans, animals or plants).
3. Mitigation of the expected climate affecting impacts of methane
emissions to the atmosphere.
At present there are, as already discussed, procedures in place that achieve the
first two of these objectives to a reasonably satisfactory level. The third,
however, is not effectively addressed at all by those approaches, and apparently
inadequately by currently used estimation methods based on EPA/GRI7.
RESULTS
An Estimate Based on Ground-Level Ambient Methane Levels
We developed a method (patent pending) to generate a preliminary estimate of
total methane emissions in Manhattan from the data collected by GSI during the
previously reported Preliminary Investigation of Ground-Level Ambient Methane
Levels in Manhattan. The method appears to be broadly applicable to other
trace gases, sites and situations. In the present case of Manhattan, such
emissions estimates can be used to assess the relative importance of those
emissions in terms of methane as a greenhouse gas (GHG) and the relative
impact of gas service/use in Manhattan in a broader climate/GHG context.
More precisely, the estimate that can be generated from the GSI Manhattan
preliminary ground-level methane data is the rate of flow of methane from
Manhattan to the atmosphere beyond.
The approach used is relatively simple. Only four pieces of information are
needed to calculate a flow rate, in this case for methane from Manhattan into
the atmosphere. What are the boundaries of the source area for the flow; in
this case what are the effective boundaries for air flow to/from Manhattan?
What is the concentration of methane in the air when the air enters the source
-------
-1 9-
area, i.e., Manhattan? What is the methane concentration when the air exits
Manhattan? How fast is the air entering/exiting Manhattan?
The GSI preliminary Manhattan methane data provide a large set of (over
700,000) measurements of the concentration of methane at various points
around the island, and other areas in the vicinity and region, at various times
over a period of five days. The challenge is to sort that data into subsets such
that the methane concentration data can be associated with air moving into
Manhattan, picking up methane in Manhattan, and then departing, and how to
estimate how much air was moving during the relevant sampling times.
Fortuitously, during certain parts of the GSI Manhattan preliminary methane
survey winds and survey pathways occurred in such patterns that evaluation of
the methane concentration in air entering and leaving Manhattan is practical. In
order to enable use of that methane data, it was necessary to gather
information and data from meteorological literature and monitoring and
reporting programs. The times and conditions of one relevant data subset from
the GSI Manhattan methane survey were as follows.
The 29 November 201 2 Methane Survey Data
From roughly 4 PM to 5 PM on the afternoon of 29 November 201 2 a survey run
was made along the west, south, and eastern sides of Lower Manhattan near
the shorelines. At that time the wind was consistent, from roughly the
southwest (compass bearing 240 degrees) at 8 miles per hour. These wind
conditions and that survey path provided data for distinct upwind and
downwind areas along the near-shoreline areas around Lower Manhattan. The
upwind data provided methane concentration of air arriving on the island, while
downwind data provided methane concentration of air departing the island on
the same wind direction path. The City College of New York has a robust
weather monitoring program. By accessing the NYCMetNet website an
estimated height for the mixing layer of the atmosphere over Manhattan for the
same time period was obtained.15 The length of the travel paths in the upwind
15 The mixing layer is the lowermost layer of air in the atmosphere where air flows over and is
influenced by the land or water surface below (see image on page 22). Above the mixing layer,
winds tend to have a smoother, laminar flow, but within the mixing layer winds tend to have
turbulent flows that cause most gases or aerosols released near the land or water surface to
disperse rapidly laterally and vertically throughout the air to the upward limit of turbulent flow.
The height of the mixing layer changes over time, but is consistent for time periods longer than
necessary for the purposes of the current data interpretation effort. Height of the mixing layer
and other meteorological data are accessible through the NYCMetNet, provided by the Optical
Remote Sensing Laboratory of The City College of New York (ORSL), http://
nvcmetnet.ccnv.cunv.edu.
-------
-20-
and downwind portions of that survey run were estimated using Google Earth.
These data were as follows:
Methane Concentrations in Ground-Level Air
Upwind 1.92 ppm ±0.003ppm (99.9999% Confidence Interval)
Downwind 2.165 ppm ±0.021 ppm (99.9999% Confidence Interval)
Wind speed (speed of air entering/exiting Manhattan) 8 mph (11.7 feet per
second)
Wind direction (from) WSW (compass bearing 240 degrees)
Manhattan wind cross-sectional length: 7 miles (36960 feet)
Mixing layer height: 2600 ft.14
These data can be applied in the following sequence of calculations:
To get the volume of air entering/leaving Manhattan per second:
Wind speed X wind cross-sectional length of Manhattan X mixing layer height =
11.7 ft/sec X 36960 ft X 2600 feet = 1.1 billion cubic feet per second
To get the amount of methane added while the air passed over Manhattan, take
the difference between the upwind and downwind methane concentrations and
apply it to the amount of air leaving Manhattan per second:
(Downwind methane concentration - Upwind concentration) X Volume of air
leaving Manhattan per second =
(2.16 ppm - 1.92 ppm) X 1,100,000,000 cu.ft./sec. = 270 cubic feet per
second
To get cubic feet per second of methane added by Manhattan to cubic feet of
methane added per year:
Cubic feet per second added by Manhattan X 60 seconds per minute X 60
minutes per hour X 24 hours per day X 365 days per year =
270 cu.ft./sec X 60 sec/min X 60 min/hr X 24 hr/day X 365 days/yr =
8,600,000,000 or 8.6 billion cubic feet per year.
This estimated annual methane flow rate from Manhattan is approximate. Each
of the measured data values used could be a source of error. The methane data
for a given time frame is highly reliable, 99.9999% confidence intervals ± <1%
(0.021 ppm). However, methane concentrations in the air vary with location,
time, wind, temperature, barometric pressure, humidity/precipitation, and the
complex collective interactions of all these and possibly other factors. To
examine the likely accuracy of the 29 November methane data used in the
above Manhattan flux estimate other data subsets from the full data set were
-------
-21 -
examined. Each of these data subsets was collected at different times, covered
different locations on and off Manhattan island, and occurred under different
weather conditions. Nevertheless each data set is still relatively large, the
smallest containing over 2000 methane data points. The following subsets
were identified and examined:
Manhattan mean methane levels relative to reference area
for given date
Date (2012)
1 1-27
1 1-29
1 1-30
12-09
Wind (from)
NE
WSW
NE
NNE
Mean Methane Concentration (ppm)
Means over
all 4 dates
Manhattan
2.079
2.165
2.345
2.261
2.21 3
Reference
Area
1.866
1.92
2.008
2.002
1.949
Increase
while over
Manhattan
Island
0.213
0.245
0.337
0.259
0.264
99.9999% Confidence interval for all Manhattan and Reference Area
Mean Methane Concentrations was < 1 % relative (0.002 to .022 ppm)
On 27 November data were collected on Manhattan island that generated a
mean methane level of 2.079 ppm, while the average methane level traveling to
NYC was 1.866 ppm. The wind that day was out of the NE (compass bearing 50
degrees) at an average speed of 5.8 mph. On this day the wind was blowing
from the area travelled to arrive in Manhattan. Hence, deducting the average
methane level before arrival in Manhattan, 1.866 ppm, from that measured in
Manhattan, 2.079 ppm, indicates the increase due to methane sources on
Manhattan island, 0.21 3 ppm. This compares reasonably well with the 0.245
ppm increase due to methane sources on Manhattan island on 29 November.
Similar data subsets were available in the 30 November and 09 December data
sets, each day with different wind conditions and, consequently, different
upwind areas used as sources of reference methane levels. On 30 November
the indicated methane concentration increase due to methane sources on
Manhattan island was 0.337 ppm. On 09 December the increase was 0.259
ppm. The table above summarizes the indicated increases in methane
-------
-22-
concentration due to sources on Manhattan island.
Given that these data subsets were for different survey paths on Manhattan,
different reference zones off the island, and on different days, times of day and
weather conditions, all effectively random, unplanned occurrences within the
body of general methane survey data, the consistency of the indicated increase
in methane concentration over Manhattan is actually impressive. In fact, the
99% confidence interval for the mean of the four days mean methane levels was
±0.068 ppm, or ±25%. Recall that the EPA/GRI7 90% confidence interval for
cast iron pipelines was ±65%. For the purposes of evaluating the likely
accuracy of the estimate of methane emissions on Manhattan Island, we will use
±25% as the likely accuracy of the data for increases in methane concentration
in the air while passing over Manhattan Island. For data quality and field
observational reasons, and to maintain a conservative approach, the 29
November data was regarded as most reliable and was used in the above
calculation of annual methane flux to the atmosphere from Manhattan.
Weather data were obtained from online sources based on National Weather
Service data or CCNY observations.15 Wind speed is likely accurate to within 0.1
mph or 0.1 5 feet per second. Winds were moderate averaging 5.5 to 6.8 mph
on the 4 survey days in the table above. The actual winds during the survey
times in the table above tended to be above the average wind speed for the
day. Since the data for 29 November was to be used in the calculation of the
Manhattan methane flux rate, the wind speed for 4PM to 5PM on that day was
estimated to be 8 mph and was the wind speed used. Potential error should
not have been greater than 10% for the wind speed used in the calculation.
Wind direction was used for two purposes. One was identification of
appropriate upwind methane reference areas and selection of an appropriate
reference data subset within the full set of methane data. The other was to
determine the length of the extent of Manhattan Island perpendicular to the
direction of the wind. This length was used because the actual volume of air
flowing over Manhattan should be related to the direction of the wind with
respect to the greater N-S length and shorter E-W width of the island. If wind
were blowing along the N-S length of the island, then, near the land surface,
the band of air blowing onto and off the island would be about 2.5 miles wide.
If the wind were blowing across the N-S length of the island, then the band of
air would be closer 10 miles wide. So, at the land surface less air would be
flowing onto and off the island for roughly N-S winds than for roughly E-W. It
might seem this would cause some difficulty in that days with N or S winds
would seem to have less air flowing over the island than days with E or W winds
of the same speed. However, the height of the mixing layer increases with time
over land compared to over water. So, this effect is probably in part
-------
-23-
compensated by related changes in the mixing layer height. In fact, on only
one (09 December) of the four days did the wind run directly along the length
of the island, and on that day the mixing layer height did increase substantially
to a height of approximately 7200 feet.15
The width of the band of air blowing over the island was the length of the
projection of the profile of Manhattan onto a line perpendicular to the wind
direction, which we call the cross-wind length. On three of the four days the
winds were nearly opposite in direction, from either the southwest or the
northeast, so the cross-wind lengths of Manhattan were very similar except on
9 December when there was a compensating increase in mixing layer height.
The cross-wind length of the island for any given wind direction can be
relatively easily estimated to within a few percent using Google Earth.
A nighttime image showing the mixing layer over Berlin, Germany. Aerosol particles dispersed
in the mixing layer cause light from below to be diffracted/'reflected revealing the mixing layer
as brighter and distinct from the clear (dark), uncontaminated air in the overlying layers of the
atmosphere. RalfSteikert http://userpage.fu-erlin.de/~kyba/images/niaht boundary layer.html
Another potential error source that might affect the calculation was the
thickness or height of the mixing layer (see image above). Equipment capable
of measuring the height of the top of the mixing layer is not common, but such
equipment is in place in Manhattan.'5 Initially, the data was obtained in a
graphic format and a 5% error was assumed due to graph reading inaccuracies.
The graphs were read conservatively to assure the height of the mixing layer
was not overestimated. The mixing layer occasionally has a somewhat diffuse
upper boundary. This occurred at 4PM-5PM on 29 November. Only the mixing
height that appeared to have the same or stronger composition (backscatter) as
near the land surface was used. This predisposes the height of the mixing layer
to underestimation as well as the resulting estimate of the actual methane flux,
-------
-24-
but, again, a conservative approach was preferred.16
Another potential source of error is the thoroughness of upward mixing of
methane in the mixing layer at the time measurements were taken in the
downwind sampling area, i.e., where air was leaving the island. Less than
thorough mixing vertically throughout the mixing layer would seem likely if
certain conditions were present. The land surface was relatively smooth, with
few tall obstructions. The gas of concern was relatively dense and diffused
slowly in the air. Winds were weak or inconsistent. The conditions during the
relevant periods of the preliminary Manhattan methane survey were the
opposite of these. Methane is lighter than air and diffuses rapidly through it,
with a tendency to move upward. Winds were appreciable and consistent. With
over 90 buildings more than 600 feet tall among many others of considerable
height (see the image of the view from the "Top of the Rock" at the beginning
of this report) the land surface of Manhattan is nearly the opposite of smooth.
Further, the graphic representations of the ceilometer data for the relevant time
periods indicated diffuse layers of air between the mixing layer and the
overlying free atmosphere. Those diffuse layers were not included in the height
of the mixing layer used in our calculations. At the time of this report, there
did not appear to be reason to assume less than thorough vertical mixing of
methane in the mixing layer. We anticipate opportunities to collect data that
more directly address this possible source of error soon, and to revise our
Manhattan methane emissions estimate in the near future.
Counter to a potential overestimate of methane emissions due to incomplete
vertical mixing of methane in the mixing layer over Manhattan, there is also an
unaccounted for potential loss of methane through the upper boundary of the
mixing layer. Methane is only about half as dense as air, and is, therefore,
strongly disposed to migrate upward in the atmosphere regardless of other
conditions. It is, therefore, likely that at any given time a portion of the
methane in the mixing layer is moving through the top of the mixing layer and
on up into the atmosphere. Such "excessive vertical mixing" would not be
accounted for in our calculations and would cause our emissions estimate to be
low. We had no data on the thoroughness of vertical mixing of methane before
the air in the mixing layer departs the island on the downwind side. We also
have no data on what proportion of methane escapes out through the top of the
mixing layer, but it seems unreasonable to expect that vertical methane loss
16 In the final stages of preparation of this report, the results of the application of two different
mixing layer algorithms to the raw ceilometer data were provided courtesy of Mark Arend and
Yonghau Wu of the City College of New York Optical Remote Sensing Lab and made available
through the NOAA CREST NYCMetNet (http://nycmetnet.ccny.cuny.edu/). The average of the
twelve results (6 time intervals X 2 algorithms) for 4PM-5PM 29 November time period was
0.81 5 kilometers, just 0.01 5 kilometers over our graphic estimate of 0.8 kilometers.
-------
-25-
would be zero. It also seems likely that either incomplete or excessive mixing
may be dominant in different areas within the downwind sampling area.
Ultimately we assumed both processes were in effect, the effects of both
countering each other in the overall data set. That is, we assumed that on
average the vertical mixing was neither incomplete nor excessive. Again, we
anticipate opportunities to collect data that will help us address this possible
source of error, and hope to release those findings, and update our emissions
estimate at the earliest practical date.
The potential error due to inadequate or excessive vertical excessive mixing in
the mixing layer could not be estimated. At the time of preparation of this
report, we had found only two publications on comparable measurement-based
methane emissions from another large metropolitan area.1718 Both were for
Krakow, Poland. The first of these, Kuc et al. (200B), estimated methane
emissions were around 760 million cubic feet per year (2.15X10 7 m3 yr_1)
over the period 1 996-1 997. The later, Zimnoch et al. (201 0), reported around
220 million cubic feet per year (6.2 X 1 0 6 m3 yr_1) over the period 2005-2009,
an apparent B.5-fold decrease from the 1996-1997 estimate. In the
intervening years the gas service operator in Krakow had undertaken a
substantial gas infrastructure improvement program, presumably substantially
reducing gas leakage. The population of Krakow is about 800,00019, while
Manhattan is very close to twice that, at 1.6 million20. The per capita gas
consumption in Poland is around 16,000 cubic feet per year21 and for New York
is around 200,000 cubic feet per year22. Adjusting the 1996-1997 Krakow
emissions for the higher population of Manhattan and New York per capita gas
consumption rate, one obtains an emissions level of 1 9 billion cubic feet per
year. The 2005-2009 Krakow emissions adjusted to Manhattan population and
NY consumption rates becomes 5.5 billion cubic feet per year. We concluded
17 T. Kuc et al. 200B. Anthropogenic emissions of CO2 and CH4 in an urban environment.
Appl. Energ. 75(3-4), 193-203.
18 M. Zimnoch et al. 201 0. Assessing surface fluxes of CO2 and CH4 in urban environment: a
reconnaissance study in Krakow, Southern Poland. Tellus (201 0), 62B, 573-580.
19 http://www.krakow-info.com/people.htm
20 http://www.nyc.gov/html/dcp/html/census/popcur.shtml
21 http://www.indexmundi.com/map/?t=Q&v=l 37000&r=eu&l=en (in cubic
meters per year per capita, converted to cubic feet per year per capita)
22 http://www.usnews.com/news/slideshows/the-l 0-states-that-use-the- least-
enerqy-per-capita/11 (in BTU per capita in 2008, converted to cubic feet per
capita per year)
-------
-26-
our estimate of 8.6 billion cubic feet per year for Manhattan is reasonable in
light of the estimates of Kuc (200B) and Zimnoch (2010) for Krakow.
In summary, among the measured data that were potential sources of error the
99% confidence interval of 25% relative for the methane concentration increase
over Manhattan was the largest likely error. Each of the other potential sources
of error were considered subject to errors of <10% relative. Further, when
interpretation of data was required, those interpretations were conservative. It
would seem reasonable at this point to hold that the estimated annual methane
flux for Manhattan may contain an error of as much as ±25%.
Comparisons of the Estimated Emissions from Manhattan
An EPA/GRI7-Factors-Based Estimate
Applying the EPA/GRI7 factors for pipe lengths and materials in the entire
ConEd system1, we arrived at an estimate of 91 5 million cubic feet as total gas
leakage from the entire ConEd system of gas mains and service connection
lines (services). Allowing an additional arbitrary 85,000,000 cubic feet for
potential leakage from other ConEd gas infrastructure, we arrived at an
estimated total methane leakage of around 1 billion cubic feet per year. Also,
because soil conditions under Manhattan probably do not support optimal
conditions for methane oxidation, we used the EPA/GRI7 methane leakage
factors instead of the methane emission factors. Use of the methane emission
factors would have generated an even lower estimate of natural gas losses/
methane emissions.
An Average Long-Term LAUF Estimate
The ConEd ten-year average of LAUF gas (reported to PHMSA) was 2.2%. Even
though the LAUF does not represent actual measured gas losses from the
ConEd system, its preparation does involve metered gas flows albeit through
many meters. Consequently, the LAUF might provide some indication of gas
losses if inherent variability can be overcome, which can be accomplished by
taking a long-term average. It should be kept in mind that 2.2% was the
average ConEd LAUF over 10 years. As the average of 10 years this value is
more reliable than the annual LAUF estimates used to calculate the average, but
this greater reliability comes with costs. The average provides a more reliable
estimate for leakage over times greater than one year, but may not be reliable
for an individual year, say, a year impacted by a major storm. Also, leak
detection and repair efforts are continuous. Use of a ten-year reporting period
in order to have a reliable leakage rate would be useless with respect to annual
-------
-27-
or more frequent efforts to identify and control leakage. For present purposes
of estimating total leakage, however, the 1 0-year average is the best value we
can extract from the reported ConEd LAUF estimates. At 2.2% the ConEd LAUF
for the entire ConEd gas system that handles about BOO billion scf/yr1 would be
6.6 billion cubic feet of lost gas, or around 6.1 billion cubic feet of methane.
The GSI Estimate Based on Preliminary Ground-Level Methane Survey Data
The actual measured levels of methane in Manhattan and adjacent areas were
used to develop an estimate of the likely rate of methane emissions from the
natural gas system in Manhattan. The estimate did not include any ConEd gas
distribution or service beyond the shorelines of Manhattan Island. The estimate
used conservative criteria in selection of which data from outside
(meteorological) sources would be used to generate the estimate. The resulting
estimate of total emissions of methane (functionally losses of natural gas) was
8.6 billion cubic feet per year («9.2 billion cubic feet of natural gas).
This estimate is 1 /3 larger than the 10-year average LAUF losses and nearly 1 0
times greater than the methane leakage estimates using the EPA/GRI7 factors
applied to the entire ConEd system of mains and services. Given that the
primary function of reported values for LAUF gas is accounting reconciliation
and equitable cost allocation, the error of 33% over the long term might be
acceptable. However, given that the 33% higher estimate was based on
methane-in-air measurements only in Manhattan, which accounts for only
about one-third of the customers and 5% of the land area in the ConEd gas
service territory, the question of how much more gas may be leaking in the
remainder of the ConEd gas system service area stands unaddressed. Similarly,
we leave for others to discuss the implications of the difference between our
estimated methane emission rate for Manhattan and the reported LAUF gas
from the entire ConEd system.
The difference between the annual Manhattan methane emission rate developed
from GSI methane survey data and that generated by application of the EPA/
GRI7 pipelines leakage factors is more striking. If one were to assume that the
EPA/GRI7 data did account for distribution and service gas lines leakage within
the accuracy given in that report (90% confidence interval was ±65% relative),
then one would would expect that the entire ConEd system might have an
emission rate up to 65% greater than the above mentioned estimate of 1 billion
cubic feet per year based on the EPA/GRI7 factors. That is, at the extreme
upper limit proposed by EPA/GRI7, the methane emissions for the entire ConEd
system should be something around 1.65 billion cubic feet per year. Even if
one uses this upper limit of an EPA/GRI7-based estimate, our estimate based on
actual methane measurements in Manhattan alone is still almost 6 times
greater.
-------
-28-
Again, a Little Bit Matters
Returning to the issue of how much methane leakage is of practical concern,
we need to put some perspective on the 8.6 billion cubic feet per year of
methane emissions that we derived from our preliminary methane data for
Manhattan. To do that we will need to make some assumptions. Our first
assumption is pipeline natural gas is 93% methane (EPA/GRI7). Our second is
that natural gas pipelines are the only sources of methane emissions on
Manhattan. Our third assumption is there are no natural gas leaks from the
ConEd system outside of Manhattan. This third assumption is obviously not
true, but allows us to put 8.6 billion cubic feet into some perspective, while
assuring that our conclusion is certainly conservative. Again, for clarification,
Manhattan comprises about only 5% of the land area and accounts for only
about 1/3 of the customers in the ConEd service territory.
Our measurements do not distinguish between methane sources. There could
be methane sources in Manhattan other than the ConEd natural gas system.
Given no data on this question at present, and based on GSI experience with
methane surveys over fairly broad areas of the Northeast, our opinion is that it
is unlikely methane from other sources would approach 10% of the emissions
level indicated by our methane survey data in Manhattan. So, for purposes of
this discussion the effects of the first two assumptions counter each other, plus
«1 0% due to 93% methane content of pipeline natural gas, and minus «1 0%
due to other potential methane sources in Manhattan.
Putting a number on the perspective for the estimated 8.6 billion cubic feet per
year methane emissions from Manhattan now requires only comparison of that
volume of gas to that handled by the ConEd system as a whole, i.e., «300
billion cubic feet per year. So our estimated annual methane emissions for
Manhattan amount to only (100 X 8.6 billion / 300 billion =) 2.86%. Once
again, why does this matter?
As mentioned back in the discussion of LAUF gas, this gas loss is actually 0.66%
greater than the long-term average ConEd LAUF of 2.2%. With respect to
hazards of explosive concentrations of methane in susceptible locations, this
amount is probably not particularly important or informative. Though it seems
reasonable to conclude such risks could increase proportionately with gas
leakage (methane emissions), that would seem to matter little as the ConEd leak
detection and management program has been running relatively effectively for
decades with no real knowledge of what actual methane emissions have been.
With respect to cost reconciliation and fair allocation, using the annual ConEd
gas sales and services revenue of 1.5 billion dollars, 0.66% is 9.9 million
dollars, consideration of which we will leave for ConEd, its customers, and
-------
-29-
NYSDPS. With respect to the impacts of methane as a greenhouse gas,
however, there is more to be said.
Methane is a potent greenhouse gas. A widely accepted minimum relative
greenhouse gas strength of methane is 21 times greater than that of carbon
dioxide over a 100-year time frame.23 There have been complex and ongoing
discussions about what the greenhouse equivalence of methane actually is,
which the reader may want to consult.24 Those discussions generally are
resulting in incremental increases in the accepted value for methane
greenhouse gas equivalence, but for this presentation we will use the simpler
approach of using the lowest widely used greenhouse equivalence for methane.
For convenience, we will further lower this by rounding it to 20 times greater
than that of carbon dioxide. So, if methane is approximately 20 times stronger
than carbon dioxide as a greenhouse gas, and if the natural gas upon reaching
its destination is entirely burned to carbon dioxide (and water), then how
important are gas (methane) leaks from the natural gas production and delivery
system that delivered it?
We can restate that methane as a greenhouse gas is 20 times stronger than
carbon dioxide by stating that it only takes 1 /20 or 5% as much methane to
cause as much atmospheric warming as a given quantity of carbon dioxide. If
the natural gas arrives at its intended destination and is burned, it will form
carbon dioxide (and water), so its original form (as methane) does not matter
since it is now carbon dioxide. However, if only 5% of natural gas escapes as it
moves from within the earth through the production, transport and delivery
systems, that 5% will have as much GHG impact as the other 95% burned as
fuel.
FINDINGS
The findings suggest the role of leakage from natural gas systems has a more
substantial role in climate change than has been appreciated.24 Apparently
present provisions in state utility regulations allow gas companies to charge
their customers for up to 2% (varies by state) of their handled gas volume as
lost and unaccounted for gas (discussed earlier in this report). Depending on
23 http://epa.gov/climatechange/ghgemissions/gases/ch4.html, or, http://www.ipcc.ch/
publications_and_data/ar4/wql /en/ch2s2-l 0-2.html, among others.
24 Alvarez, R. A., Pacala, S. W. Winebrake, J. J., Chameides, W. L. & Hamburg, S. P. Greater focus
needed on methane leakage from natural gas infrastructure. Proc. Natl Acad. Sci. USA 1 09,
6435-6440 (2012).
-------
-30-
the state, presumably such allowances apply to each sector of the gas system
separately, i.e., production (gas wells), transportation (long distance pipelines),
and distribution (gas utilities). In the end the methane emissions that affect the
greenhouse gas impact of natural gas as fuel are the total methane emissions
along the whole path the gas travels through the entire production-transport-
distribution network. The infrastructure in each sector in that network can and
does leak natural gas.
A 2.86% leakage of all the natural gas handled by ConEd in Manhattan alone
leaves only 2.1 4% for the rest of the ConEd system, and the production and
transport system feeding it, to leak collectively before total losses exceed the
5% level at which the greenhouse gas cost of using natural gas is effectively at
least doubled. So far GSI efforts to gather data on volumes of gas lost by
leakage or other processes in the natural gas system have indicated all such
data are based on methods that are not founded in well-documented data on
actual leaks, let alone actual measurements of leaks or field emissions. Some
actual field data have recently been reported for production and early stage
transport of shale gas. In the Denver-Julesberg Fossil Fuel Formation, largely in
Weld County in northeast Colorado, emissions of methane were estimated at
2.3% to 7.7% of production.25 Preliminary results from the Uinta Basin in Utah
discussed at recent meetings of the American Geophysical Union indicated
methane leakage in the field reached 9% of total production.26 Even if the
Marcellus shale gas fields planned to serve New York City release methane
emissions at the lowest rate indicated by field data from northeast Colorado,
and if that were added to just the GSI estimated methane emission for
Manhattan alone, that would already put the total methane emission leak rate
for Marcellus Shale gas delivered through the ConEd system at 5.16%. This
leakage rate, which does not account for leakage from gas transmission lines to
ConEd or from the rest of the ConEd system outside Manhattan, is already in
excess of our simple calculation for the total leakage rate (5%) at which the
leaked gas has as much potential climate impact as the burned gas. In fact,
this leakage is well in excess of the total leakage rate of 3.2% at which other
authors using more elaborate approaches have concluded that natural gas
ceases to have a "clean fuel" advantage over coal for power production.18
25 Gabriel Petron et al. Hydrocarbon Emissions Characterization in the Colorado Front Range -
A Pilot Study. National Oceanic and Atmospheric Administration, Earth System Research
Laboratory, Boulder, Colorado, USA. (Nature 482, 1 3 9 —=1 4 0 ; 2012)
26 http://www.nature.com/news/methane-leaks-erode-green-credentials-of-natural-
gas-1.1 21 23#/ref-link-4
-------
-31 -
Caveats and Cautions Regarding the GSI Preliminary Estimate of the Manhattan
Methane Emissions Rate
The GSI method (patent pending) used to estimate the Manhattan methane
emissions rate from preliminary mobile methane survey data does not provide
an estimate that is relative to natural background levels for natural areas in the
region. It is difficult to imagine that there might even be an area anywhere in
the vicinity of New York City where natural background methane emissions
rates might be evaluated. The GSI approach was instead based on an alternate
approach that could be evaluated because Manhattan is an island making
physical boundaries of the Manhattan land surface emissions area relatively
easy to define. Further, because of observations during the methane survey
and analyses of the survey data, it became apparent that air arriving on the
upwind and departing the downwind sides of the island at any given time
necessarily provide a functional methane baseline and impacted air
concentration level for the island. Hence, it is not necessary to know the
natural methane baseline for the area or region, or even the surrounding
waters, in order to calculate an emission rate for the island. Also, this approach
eliminates any need to understand or attempt to correct off-island incoming air
methane concentrations for methane sources within the geographical methane
reference area since the only needed data is methane concentration in the
incoming air.
The height of the mixing layer is important to the accuracy of the GSI approach
to estimating area methane emissions based on ground level methane
concentrations. Fortunately mixing height data is measured in Manhattan.
However, the measurement used was collected at a single location not in the
area where the departing air methane concentration data were collected.
Nevertheless due to the mixing layer measurement location being relatively
upwind from the air departure area it is more likely the mixing layer height
used was too low rather than too high. Also, the measurement used was
chosen to exclude diffuse zones at the upper edge of the mixing layer. Actual
above ground and airborne measurements would be useful to assess variations
of concentration of methane throughout the mixing layer.
There are potential and actual sources of methane in Manhattan other than the
ConEd natural gas system. The GSI approach to estimating methane emissions
cannot distinguish the contributions of various potential sources of methane to
the overall methane emissions rate. One clearly distinguishable localized
release of possible "sewer gas" was observed in the GSI Manhattan methane
survey data collected at the outlet of a storm drain on the east side of the
island. The elevated methane level was apparent, but not particularly high.
How many other methane elevations might have been due to sewer gas or other
-------
-32-
potential, non-ConEd, methane sources, e.g., old fill areas, is not known.
However, based on GSI experience in other urban and rural areas, the effects of
using conservative allowances and assumptions wherever reasonable likely
exceed the influence of landfill, sewer or other biologically generated methane
in the GSI Manhattan preliminary methane emissions estimate. The relative
importance of biogenic methane sources in Manhattan probably could be
assessed using methane isotopic composition analysis. It is also worthwhile to
note that just because gas is being released from a sewer or storm drain does
not necessarily confirm that the gas is actually generated in the sewage or
storm water and residues. Sewers and storm drains can also receive and
transport gas leaked from gas pipes.
There is also potential for losses due to pirated or illegal gas taps, and post-
metering losses at the consumer level. Again, such losses cannot be
distinguished within the GSI Manhattan methane emissions estimate, but seem
likely to be small in comparison to leakage from ConEd gas infrastructure and
operations.
RECOMMENDATIONS
The estimated Manhattan methane emission rate presented in this report
indicates the need for actual measurements of methane flux for urban,
petroleum and gas field areas, etc. instead of estimates based on
extrapolations of typically very limited and generally indirect data.
In Manhattan, additional ground level methane survey work seems needed to
support more effective and rapid detection and identification of gas leaks, to
determine areas where gas pipe is in need of general replacement or lining
rather than stop-gap repairs. Additional ground level work is needed that is
specifically designed to develop and refine the approach developed and
presented in this report for rapid actual-measurement-based estimation of
methane emissions. Additional supplementary work is needed to explore and
refine the level of knowledge regarding the height of the mixing layer and
methane distribution within it for Manhattan and other urban and non-urban
settings.
The findings from this data analysis effort indicate there is need to re-evaluate:
• Methane emissions estimates and assumptions being used as the basis for
global climate modeling and projections regarding the path and speed of
climate change
• Plans and projections regarding short-term high-impact opportunities to
reduce greenhouse gas emissions by focusing initially on methane emissions
-------
-33-
associated with fossil fuel development, production, and utilization
• Regulation of the fossil fuel industry
• The actual economic and environmental costs of fossil fuel compared to
alternative energy technologies over all time frames.
Our findings, based on actual measurements, necessarily raise doubts about
the claimed value of natural gas as a "clean, bridge fuel" and call for further
work to verify the reported findings and to begin to identify specific methane
sources and improve natural gas leak prevention and management.
ACKNOWLEDGEMENTS
We want to express our appreciation to the scientists and other colleagues who
reviewed drafts and provided comments and suggestions during preparation of
this report.
-------
Appendix D
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
a
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
ENP06776_proof ¦ 16 November 2012 ¦ 1/4
Environmental Pollution xxx (2012) 1e4
ELSEVIER
Contents lists available at SciVerse ScienceDirect
Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol
ENVIRONMENTAL
POLLUTION
Rapid communication
Mapping urban pipeline leaks: Methane leaks across Boston
qi Nathan G. Phillips3 *, Robert Ackleyb, Eric R. Crossonc, Adrian Downd, Lucy R. Hutyra3, Max Brondfield3,
Jonathan D. Karrd, Kaiguang Zhaod, Robert B. Jacksond
a Boston University, Department of Earth and Environment, 675 Commonwealth Avenue, Boston, MA 02215, USA
bGas Safety, Inc., South borough, MA 01772, USA
cPicarro, Inc., Santa Clara, CA 95054, USA
dDuke University, Nicholas School of the Environment and Center on Global Change, Durham, NC 27708, USA
article info
Article history:
Received 25 July 2012
Received in revised form
31 October 2012
Accepted 3 November 2012
eevwords: .
arbon isotopes
Infrastructure
Methane
Natural gas
Urban
abstract
Natural gas is the largest source of anthropogenic emissions of methane (CH4) in the United States. To
assess pipeline emissions across a major city, we mapped CH4 leaks across all 785 road miles in the city of
Boston using a cavity-ring-down mobile CH4 analyzer. We identified 3356 CH4 leaks,with concentrations
exceeding up to 15 times the global background level. Separately, we measured d CH4 isotopic signa-
tures from a subset of these leaks. The d13CH4 signatures (mean Vi- 42.8& ± 1.3& s.e.; n % 32) strongly
indicate a fossil fuel source rather than a biogenic source for most of the leaks; natural gas sampled
across the city had average d13CH4 values of-36.8& (±0.7& s.e., n % 10), whereas CH4 collected from
landfill sites, wetlands, and sewer systems had d13CH4 signatures w20& lighter (m % -57.8&, ±1.6& s.e.,
n Zi 8). Repairing leaky natural gas distribution systems will reduce greenhouse gas emissions, increase
consumer health and safety, and save money.
© 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Methane (CH4) is a greenhouse gas more potent molecule for
molecule than carbon dioxide (Shindell et al., 2012). In the United
States, leaks of CH4 from natural gas extraction and pipeline
transmission are the largest human-derived source of emissions
(EPA, 2012). However, CH4 is not just a potent greenhouse gas; it
also influences air quality and consumer health. CH4 reacts with
NOxto catalyze ozone formation in urban areas (West etal., 2006).
Incidents involving transmission and distribution pipelines for
natural gas in the U. S. cause an average of 17 fatalities, 68 injuries,
and $133 M in property damage each year (PHMSA, 2012). A natural
gas pipeline explosion in San Bruno, CA, for instance, killed eight
people and destroyed 38 homes in 2010. Detecting and reducing
pipeline leaks of CH4 and other hydrocarbons in natural gas are
critical for reducing greenhouse gas emissions, improving air
quality and consumer safety, and saving consumers money (V\fest
et al., 2006; Han and V\feng, 2011; Shindell et al., 2012; Alvarez
etal., 2012).
To assess CH4 emissions in a major urban metropolis, we map-
ped CH4 emissions over the entire 785 centerline miles of Boston's
* Corresponding author.
E-mail address: nathan@bu.edu (N.G. Phillips).
0269-7491/$ e see front matter © 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.Org/10.1016/j.envpol.2012.11.003
streets. To evaluate the likely source of the street-level CH4 emis-
sions, we also measured the d13CeCH4 carbon isotope composition,
which can differentiate between biogenic (e.g., landfill, wetland,
sewer)andthermogenic(e.g., natural gas)sources(Schoell, 1980).
2. Materials and methods
We conducted 31 mobile surveys during the period 18 August, 2011 e1 October,
2011, covering all 785 road miles within Boston's city limits. We measured CH4
concentration ([CH4], ppm) using a mobile Picarro G2301 Cavity Ring-Down Spec-
trometer equipped with an A0491 Mobile Plume Mapping Kit (Picarro, Inc, Santa
Clara, CA). This instrument was factory-calibrated on 15 August 2011, immediately
prior to use in this study, and follow-up tests of the analyzer were made during 11 e
21 August, 2012, comparing analyzer output to a National Oceanic and Atmospheric
Administration (NOAA) primary standard tank. In both pre- and post-checks, the
analyzer output was found to be within 2.7 parts per billion of known [CH4] in
standard tanks, three orders of magnitude below typical atmospheric concentra-
tions. Spectrometer and mobile GPS data were recorded every 1.1 s. To correct for
a short time lag between instantaneous GPS location and a delay in [CH4]
measurement due to inlet tube length (w3 m), we used an auxiliary pump to
increase tubing flow throughput to within 5 cm of the analyzer inlet; we also
adjusted the time stamp on the [CH4] readings based on a 1-s delay observed
between analyzer response to a standard CH4 source that we injected into the
instrument while driving, and the apparent GPS location. We also checked the GPS-
based locations of leaks with dozens of street-level sampling to confirm specific leak
locations and the estimated sampling delay. Air was sampled through a 3.0 um
Zefluor filter and Teflon tubing placed w30 cm above road surfaces.
For our mobile survey data, we defined a "leak" as a unique, spatially contiguous
group of [CH4] observations, all values of which exceed a concentration threshold of
2.50 ppm. This was used as a threshold because it corresponded to the 90th
Please cite this article in press as: Phillips, N.G., etal., Mapping urban pipeline leaks: Methane leaks across Boston, Environmental Pollution
(2012), http://dx.doi.Org/10.1016/j.envpol.2012.11.003
-------
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
ENP06776_proof ¦ 16 November 2012 ¦ 2/4
N.G. Phillips et al. / Environmental Pollution xxx (2012) 1e4
percentile of the distribution of data from all road miles driven, and, relative to
global background, is w37% above 2011 mean mixing ratios observed at Mauna Loa
(NOAA, 2012).
Independently of mobile street sampling of CH4, we measured d13CH4 from
a subset of the leaks with a PicarroG2112i Cavity Ring-Down Spectrometer (Crosson,
2008). This instrument is calibrated monthly using isotopic standards from
Isometric Instruments (Victoria, BC, Canada). The instrument was checked at least
once daily to ensure analyzer output was within 1& of a tank of CH4 with d13CH4
measured by a private lab (Isotech Labs, IL). Samples were collected in 1-L Tedlar
sampling bags with valve and septa fittings, manufactured by Environmental Supply
Company (Durham, NC). A Gas Sentry CGO-321 handheld gas detector (Bascom-
Turner, MA) was used to identify the area of highest ambient [CH4] at each site
sampled for d13CH4. Sampling bags were pre-evacuated and filled at the area of
highest ambient concentration at the sampling site using a hand pump, d 13CH4 was
analyzed using a Picarro G2112i with a sample hold time typically of a few days and
always less than two weeks.
At a subset of sampling sites (n % 12), we collected duplicate samples in glass
vials to assess potential leaking or fractionation by the Tedlar sampling bags. We also
sent duplicate samples from a different subset of sampling sites (n % 5) to a private
lab (Isotech Labs, IL) for independent d CH4 analysis. These analyses suggest no
significant fractionation or bias either from the sampling bags or the Picarro G2112i
analyzer. Most samples were analyzed at less than the maximum hold time of two
weeks, at which bag diffusion could account for a 1.2& drift in our measurements of
d13CH4.
We compared d13CH4 of these locations with samples taken from area landfills,
wetlands, and the Deer Island Water Treatment Facility. Sampling equipment and
procedures, as well as laboratory analyses, for landfill and wetland sites were similar
to those ford CH4 sampling locations described above. Samples were collectedfrom
three capped, inactive landfills (there are currently no active landfills in the Boston
area). At one former landfill site, samples were collected at approximately three-
month intervals between September, 2011 and April, 2012. The d13CH4 signature
of the landfill was consistent over this period (±3.4& s.e.). At all wetland sampling
sites, a plastic chamber (10 cm x 25 cm x 5 cm) connected to a sampling tube was
placed over the surface of exposed moist sediment or shallow (>5 cm) water.
Sediment below the chamber was disturbed gently before drawing air samples from
the headspace within the chamber. The sample from the Deer Island Treatment
Facility was drawn from the headspace of a sample bottle of anaerobic sludge,
collected onsite by Deer Island staff for daily monitoring of the facility's anaerobic
sludge digesters.
3. Results arid discussion
V\fe identified 3356 CH4 leaks (Figs. 1 and 2) exceeding 2.50 parts
per million. Surface concentrations corresponding to these leaks
ranged up to 28.6 ppm, 14-times above a surface background
concentration of 2,07 ppm (the statistical mode of the entire
concentration distribution). Across the city, 435 and 97 indepen-
dent leaks exceeded 5 and 10 ppm, respectively.
Based on their d13CH4 signatures, the CH4 leaks strongly
resembled thermogenic rather than biogenic sources (Fig. 3).
Samples of natural gas from the gateway pipelines to Boston and
from other consumer outlets in the city were statistically indis-
tinguishable, with an average d 'CH.s signature of -36.8& (±0.7&
s.e., n % 10; & vs. Vienna Pee Dee Belemnite). In contrast, CH4
collectedfrom landfill sites, wetlands, and sewer systems reflected
a greater fractionation from microbial activity and d13CH4 signa-
tures w20& lighter. Biogenic values ranged from -53.1 &
to -64.5& (m % -57.8&, ±1.6& s.e., n % 8) for samples collected in
four wetlands, three capped landfills, and the primary sewage
facility for the city, Deer Island Sewage Treatment Plant, which had
the heaviest sample observed for non-natural-gas sources
(-53.1 &). Our results for biogenic CH4 carbon isotope signatures
are consistent with other studies of the d13CH4 signature of CH4
from landfills (Bergamaschi et al., 1998; Borjesson et al., 2001) and
wetlands (Hornibrook et al., 2000).
Peaks of [CH4] detected in the road surveys strongly reflected
the signature of natural gas rather than biogenic sources (Table 1).
The average d l3CH4 value for peaks was -42.8& ± 1.3& (n % 32),
reflecting a dominant signal from natural gas, likely altered in some
cases by minor fractionation of natural gas traveling through soils
and by mixing with background air (d13CH4 % -47&; Dlugokencky
et al.. 2011). A minority of samples had d i3CH4 more negative than
Fig. 1. Upper Panel: Methane leaks (3356 yellow spikes > 2.5 ppm) mapped on
Boston's 785 road miles (red) surveyed in this study. Lower Panel: Leaks around
Beacon Hill and the Massachusetts State House. Sample values of methane concen-
trations (ppm) are shown for each panel. (For interpretation of the references to color
in this figure legend, the reader is referred to the web version of this article.)
that of background air, reflecting apparent influence of biogenic
CH4. All 32 samples emitted a distinct odor of the mercaptan
additive associated with natural ggs, including those with a larger
apparent biogenic influence on d CH4.
o 900
o
_c
f> 600
CD
Z
co 300
CO
a>
-1 0
0 50 100 150
Miles of Cast Iron Mains
per Neighborhood
Fig. 2. Leak prevalence is associated with old cast iron pipes across ten Boston
neighborhoods. (The combined line is the regression across all ten neighborhoods
(P < 0.001); the green regression line [r2 1/4 0.34; P Va 0.08], which eliminates the
influence of the leverage point [Dorchester neighborhood], has a slope and intercept
indistinguishable (P > 0.10) from the combined regression.). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this
article.)
Boston, MA
Please cite this article in press as: Phillips, N.G., et al., Mapping urban pipeline leaks: Methane leaks across Boston, Environmental Pollution
(2012), http://dx.doi.Org/10.1016/j.envpol.2012.11.003
-------
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
ENP06776_proof ¦ 16 November 2012 ¦ 3/4
N.G. Phillips et al. / Environmental Pollution xxx (2012) 1e4
CM - -—
-60 -50 -40 -30
S"C (%o PDB)
-20
Fig. 3. d CH4 of [CH4] peaks detected in road surveys (n Va 32). Red lines represent
means of thermogenic (-36.8&, ±0.7& s.e., n 1/4 10) and biogenic (-57.8S, ±1.6& s.e.,
n 1/4 8) sources, respectively. (For interpretation of the references to color in this figure
legend, the reader is referred to the web version of this article.)
Leaks across Boston (Fig. 1), were associated primarily with cast
iron mains that were sometimes over a century old (Fig. 2). Across
ten Boston neighborhoods, leak frequency was linearly related to
number of miles of cast iron mains (r2 % 0.79, P < 0.001; Fig. 2), but
only marginally to miles of non-cast-iron piping (r2 %0.27; P %0.12,
data not shown). Leak counts did not differ statistically by neigh-
borhood or by socio-economic indicators for the neighborhoods
Table 1
Locations and isotopic values from discrete street leak samples.
Latitude
Longitude
d13CH4(&PDB)
42.3654
-71.0612
-53.959
42.3439
-71.2628
-47.898
42.3493
-71.2265
-57.590
42.3583
-71.1749
-40.818
42.3411
-71.2440
-37.323
42.3543
-71.2441
-38.241
42.3559
-71.1898
-39.412
42.3513
-71.2092
-41.978
42.3515
-71.2081
-39.531
42.3614
-71.2314
-41.796
42.3426
-71.1012
-44.100
42.3443
-71.0949
-41.848
42.3328
-71.0761
-37.516
42.3360
-71.0738
-46.414
42.3441
-71.0673
-45.490
42.3303
-71.0569
-37.476
42.3409
-71.0542
-40.029
42.3524
-71.0445
-43.127
42.3799
-71.0272
-48.182
42.3722
-71.0361
-57.693
42.3785
-71.0681
-48.429
42.3730
-71.0632
-37.471
42.3593
-71.0629
-42.689
42.3584
-71.0644
-52.033
42.3546
-71.1271
-47.241
42.2943
-71.1891
-52.028
42.2793
-71.1514
-37.648
42.2887
-71.1428
-32.467
42.3285
-71.0792
-28.251
42.3215
-71.0692
-36.214
42.3269
-71.0796
-30.662
42.3553
-71.0573
-43.836
Mean
-42.793
Standard error
1.259
obtained from the 2010 US Census (P > 0.1 for number of housing
units and ethnicity) or the 2000 US Census (P > 0.1 for median
income and poverty rate).
Reducing CH4 leaks will promote safety and help save money.
Although our study was not intended to assess explosion risks, we
observed six locations where gas concentrations in manholes
exceeded an explosion threshold of 4% [CH4] at 20 °C (concentra-
tions measured using a Gas Sentry CGO-321 handheld gas detector;
Bascom-Turner, MA). Moreover, because CH4, ethane (C2H6), and
propane (C3H8) interact with NOx to catalyze ozone formation,
reducing these hydrocarbon concentrations should help reduce
urban ozone concentrations and respiratory and cardiopulmonary
disease (West et al., 2006; Shindell et al., 2012). CH4 is also a potent
greenhouse gas, with an estimated 20-year global warming
potential 72 times greaterthan CO2 (Alvarez et al., 2012; Townsend-
Small et al., 2012). Replacing failing natural gas mains will reduce
greenhouse gas emissions, thereby providing an additional benefit
to the fewer mercury, SO2 and particulate emissions that natural-
gas burning emits compared to coal (Shindell et al., 2012). Finally,
leaks contribute to $3.1 B of lost and unaccounted natural gas
annually in the United States (EIA, 2012; 2005e2010 average).
Our ongoing and future research evaluates how surface [CH4]
values correspond to individual, and city-wide, urban leak rates and
greenhouse-gas emissions. Two approaches to this question are
useful: "bottom-up" chamber measurements taken on represen-
tative samples of individual leaks, and "top-down" atmospheric
mass-balance estimates of the collective urban leak rate that
exploit the known isotopic signature of natural gas versus that of
biogenicsources and otherfossil fuel sources. The instrumentation
used in this study is well-suited for both approaches.
We propose that a coordinated campaign to map urban pipeline
leaks around the world would benefit diverse stakeholders,
including companies, municipalities, and consumers. Repairing the
leaks will bring economic, environmental, and health benefits to all.
Acknowledgments
The Barr Foundation, Conservation Law Foundation, Picarro, Inc.,
Duke University's Center on Global Change and Nicholas School of
the Environment, and Boston University's Sustainable Neighbor-
hood Laboratory supported this research. Dr. Michael Delaney of
the Massachusetts V\feter Resources Agency facilitated sewage
influent sampling. Additional support was provided by the US
National Science Foundation ULTRA-ex program (DEB 0948857).
Shanna Cleveland, Adrien Finzi and Steven V\fofsy provided helpful
comments on the manuscript.
References
Alvarez, R.A., Pacala, S.W., Winebrake, J.J., Chameides, W.L., Hamburg, S.P, 2012.
Greater focus needed on methane leakage from natural gas infrastructure.
Proceedings of the National Academy of Sciences U.S.A. 109, 6435e6440.
Bergamaschi, P., Lubina, C., Knigstedt, R., Fischer, H., Veltkamp, A.C., Zwaagstra, O.,
1998. Stable isotopic signatures (d13C, dD) of methane from European landfill
sites. Journal of Geophysical Research 103, 8251e8265.
Borjesson, G., Chanton, J., Svensson, B.H., 2001. Methane oxidation in two Swedish
landfill covers measured with carbon-13 to carbon-12 isotope ratios. Journal of
Environmental Quality 30, 369e376.
Crosson, E.R., 2008. A cavity ring-down analyzer for measuring atmospheric levels
of methane, carbon dioxide, and water vapor. Applied Physics B: Lasers and
Optics 3, 403e408.
Dlugokencky, E.J., Nisbet, E.G., Fisher, R., Lowry, D., 2011. Global atmospheric
methane: budget, changes and dangers. Philosophical Transactions of the Royal
Society A 369, 2058e2072.
Energy Information Administration (EIA), 2012. http://205.254.135.7/dnav/ng/ng_
sum_lsum_dcu_nus_a.htm, http://205.254.135.7/naturalgas/annual/pdf/table_
a01.pdf, htt p://205.254.135.7/natura I gas/a nnua l/arc hive/2009/pdf/table_a01.
pdf.
Environmental Protection Agency (EPA), 2012. http://epa.gov/methane/sources.
html.
Please cite this article in press as: Phillips, N.G., etal., Mapping urban pipeline leaks: Methane leaks across Boston, Environmental Pollution
(2012), http://dx.doi.Org/10.1016/j.envpol.2012.11.003
-------
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
ENP06776_proof ¦ 16 November 2012 ¦ 4/4
N.G. Phillips et al. / Environmental Pollution xxx (2012) 1e4
4
Han, Z.Y., Weng, W.G., 2011. Comparison study on qualitative and quantitative risk
assessment methods for urban natural gas pipeline network. Journal of
Hazardous Materials 189, 509e518.
Hornibrook, E.R.C., Longstaffe, F.J., Fyfe, W.F., 2000. Evolution of stable carbon
isotope compositions for methane and carbon dioxide in freshwater wetlands
and other anaerobic environments. Geochimica Cosmochimica Acta 64, 1013e
1027.
National Oceanic and Atmospheric Administration, Annual Greenhouse Gas Index
(AGGI), 2012. Earth System Research Laboratory, Global Monitoring Division.
http://www.esrl. noaa.gov/gmd/aggi/(accessed 10.09.12.).
Pipeline and Hazardous Materials Safety Administration (PHMSA), 2012. www.
phmsa.dot.gov/pipeline/library/data-stats.
Schoell, M., 1980. The hydrogen and carbon isotopic composition of methane from
natural gases of various origins. Geochimica Cosmochimica Acta 44, 649e661.
Shindell, D., Kuylenstierna, J.C.I., Vignati, E., van Dingenen, R., Amann, M.,
Klimont, Z., Anenberg, S.C., Muller, N., Janssens-Maenhout, G., Raes, R,
Schwartz, J., Faluvegi, G., Pozzoli, L., Kupiainen, K., Hoglund-lsaksson, L.,
Emberson, L., Streets, D., Ramanathan, V., Hicks, K., Oanh, N.T.K., Milly, G.,
Williams, M., Demkine, V., Fowler, D., 2012. Simultaneously mitigating near-
term climate change and improving human health and food security. Science
335, 183e189.
Townsend-Small, A., Tyler, S.C., Pataki, D.E., Xu, X., Christensen, L.E., 2012. Isotopic
measurements of atmospheric methane in Los Angeles, California, USA: influ-
ence of "fugitive" fossil fuel emissions. Journal of Geophysical Research 117,
D07308.
West, J.J., Fiore, A.M., Horowitz, L.W., Mauzerall, D.L., 2006. Global health benefits of
mitigating ozone pollution with methane emission controls. Proceedings of the
National Academy of Sciences U.S.A. 103, 3988e3993.
Please cite this article in press as: Phillips, N.G., etal., Mapping urban pipeline leaks: Methane leaks across Boston, Environmental Pollution
(2012), http://dx.doi.Org/10.1016/j.envpol.2012.11.003
-------
Appendix E
-------
JOURNAL OF GEOPHYSICAL RESEARCH: ATMOSPHERES, VOL. 118,4974^1990, doi:10.1002/jgrd.50413, 2013
Quantifying sources of methane using light alkanes
in the Los Angeles basin, California
J. Peischl,1'2 T. B. Ryerson,2 J. Brioude,1'2 K. C. Aikin,1'2 A. E. Andrews,3 E. Atlas,4
D. Blake,5 B. C. Daube,6 J. A. de Gouw,1'2 E. Dlugokencky,3 G. J. Frost,1'2
D. R. Gentner,7 J. B. Gilman,1'2 A. H. Goldstein,7'8 R. A. Harley,7 J. S. Holloway,1'2
J. Kofler,1'3 W. C. Kuster,1'2 P. M. Lang,3 P. C. Novelli,3 G. W. Santoni,6 M. Trainer,2
S. C. Wofsy,6 and D. D. Parrish2
Received 26 November 2012; revised 10 April 2013; accepted 12 April 2013; published 28 May 2013.
[i] Methane (CH4), carbon dioxide (C02), carbon monoxide (CO), and C2-C5 alkanes
were measured throughout the Los Angeles (L.A.) basin in May and June 2010. We use
these data to show that the emission ratios of CH4/CO and CH4/CO2 in the L.A. basin
are larger than expected from population-apportioned bottom-up state inventories,
consistent with previously published work. We use experimentally determined CH4/CO
and CH4/CO2 emission ratios in combination with annual State of California CO and C02
inventories to derive a yearly emission rate of CH4 to the L.A. basin. We further use the
airborne measurements to directly derive CH4 emission rates from dairy operations in
Chino, and from the two largest landfills in the L.A. basin, and show these sources are
accurately represented in the California Air Resources Board greenhouse gas inventory for
CH4. We then use measurements of C2-C5 alkanes to quantify the relative contribution of
other CH4 sources in the L.A. basin, with results differing from those of previous studies.
The atmospheric data are consistent with the majority of CH4 emissions in the region
coming from fugitive losses from natural gas in pipelines and urban distribution systems
and/or geologic seeps, as well as landfills and dairies. The local oil and gas industry also
provides a significant source of CH4 in the area. The addition of CH4 emissions from
natural gas pipelines and urban distribution systems and/or geologic seeps and from the
local oil and gas industry is sufficient to account for the differences between the top-down
and bottom-up CH4 inventories identified in previously published work.
Citation: Peischl, J., et al. (2013), Quantifying sources of methane using light alkanes in the Los Angeles basin,
California, J. Geophys. Res. Atmos., 118, 4974^1990, doi:10.1002/jgrd.50413.
1. Introduction
[2] In California, methane (CH4) emissions are regulated
by Assembly Bill 32, enacted into law as the California
Global Warming Solutions Act of 2006, requiring the
state's greenhouse gas (GHG) emissions in the year 2020
not to exceed 1990 emission levels. To this end, the California
Air Resources Board (CARB) was tasked with compiling and
verifying an inventory of GHG emissions for the state. Two
published works [Wunch et al., 2009; Hsu et al., 2010]
have concluded that atmospheric emissions of CH4 in the
'Cooperative Institute for Research in Environmental Sciences,
University of Colorado Boulder, Boulder, Colorado, USA.
2Chemical Sciences Division, National Oceanic and Atmospheric
Administration Earth System Research Laboratory, Boulder, Colorado,
USA.
Corresponding author: J. Peischl, Chemical Sciences Division, National
Oceanic and Atmospheric Administration Earth System Research Labora-
tory, Boulder, CO, USA. (jeff.peischl@noaa.gov)
©2013. American Geophysical Union. All Rights Reserved.
2169-897X/13/10.1002/jgrd.50413
Los Angeles (L.A.) area were greater than expected from a
per capita apportionment of the statewide 2006 CARB
GHG inventory and from a bottom-up accounting of CH4
sources, respectively.
[3] Several recent works have estimated CH4 emissions to
the South Coast Air Basin (SoCAB; Figure la), which are
summarized in Table 1. Wunch et al. [2009] used a Fourier
transform infrared spectrometer at the Jet Propulsion
Laboratory (JPL) in Pasadena, California to measure
vertically integrated total column enhancement ratios of
3Global Monitoring Division, National Oceanic and Atmospheric
Administration Earth System Research Laboratory, Boulder, Colorado, USA.
4Rosenstiel School of Marine and Atmospheric Science, University of
Miami, Miami, Florida, USA.
'Department of Chemistry, University of California, Irvine, California,
USA.
^School of Engineering and Applied Science and Department of Earth and
Planetary Sciences, Harvard University, Cambridge, Massachusetts, USA.
'Department of Civil and Environmental Engineering, University of
California, Berkeley, California, USA.
sDepartment of Environmental Science, Policy, and Management,
University of California, Berkeley, California, USA.
4974
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
A MWO
O Caltech
NASA JPL
highways
O landfills
O dairies
10
n
longitude, 5W
0~~
k
"i
V,
i
Pit ^
w *|f *
I
I
1850
8000
-7000
6000
5000
-4000
B
-3000
2000
1000
8000
- 6000
4000
- 2000
118.0 117.5
longitude, "W
Figure 1. (a) Map of California. The dashed box shows the inset for Figure lb; the solid box shows
the extent of the map boundaries for Figures lc-le. (b) Map of southern California showing the
location of downtown L.A. (blue dot), the Los Angeles County boundary (green), the South Coast
Air Basin boundary (red), and the extent of the map boundaries for Figures lc-le (black box), (c) Map
of the L.A. region showing known sources of CH4 in the L.A. basin. The white triangle shows the loca-
tion of the Mount Wilson Observatory (MWO), where ground-based measurements were made by Hsu
et al. [2010] and in this study. The light blue star shows the location of the Jet Propulsion Laboratory,
where Wimch et al. [2009] made their measurements. The California Research at the Nexus of Air Quality
and Climate Change (CalNex) Pasadena ground site was located on the California Institute of Technology
(Caltech) campus, located at the orange-filled circle. Landfills (white circles) and CH4 point sources
(filled blue circles; negligibly small) are sized by emissions in the 2008 CARB greenhouse gas inventory.
Dairies (filled yellow circles) are sized by the estimated emissions from the number of cows from Salas
et al. [2008] multiplied by the 2009 CARB GHG inventory annual CH4 emission per cow from enteric
fermentation, (d) Same map of the Los Angeles region as in Figure lc, with flight tracks from 16 daytime
flights of the NOAA P-3 (thin black lines). CH4 measurements from the daytime boundary layer are
color-coded atop these tracks according to the legend to the right, (e) Locations of whole air samples
in the L.A. basin are colored by ethane mixing ratio and sized by propane mixing ratio as indicated in
the legends to the right. JPL, Jet Propulsion Laboratory.
4975
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
Table 1. Summary of Past Studies Investigating CH4 Emissions in the L.A. Basin
Percentage of California
CH,
Bottom-up CH4
Geographic
Population in
Emission
Emission Inventory
Study
Time of Study
Area
Geographic Area
(Gg/yr)
Inventory Referenced
(Gg/yr)
Wunch et al. [2009]
August 2007 to June 2008
SoCAB
43%
400±100
600±100
CARB CO 2007
(CARB C02
2006 +EDGAR C02
2005)/2
260b
Hsu etal. [2010]
April 2007 to May 2008
L.A. County
0 SoCAB
27%
200 ± 10
CARB CO 2007
140
Wennberg et al. [2012]
April 2007 to May 2008
SoCAB
43%
380a± 100
CARB CO 2007
-
June 2008
SoCAB
43%
470±100
CARB CO 2008
-
Mav 2010 to June 2010
SoCAB
43%
440 ± 100
CARB CO 2010
-
'Wennberg et al. [2012] recalculated the data reported by Hsu et al. [2010] to estimate a CH4 emission from the entire SoCAB.
bWunch et al. [2009] apportioned the statewide CARB GHG inventory for CH4, less agriculture, and forestry emissions, by population.
CH4 relative to CO and to CO2. The observed column
enhancement ratios, multiplied by CARB inventory values
of CO for 2008 and an average of 2006 CARB GHG
inventory and 2005 Emission Database for Global Atmospheric
Research (EDGAR) for CO2, were used to derive a lower
limit to CH4 emissions of 400 ± 100 Gg CH i/yr (based on
CO) or 600 ± 100 Gg CH4/yr (based on CO2) for the
SoCAB. One reason for the discrepancy in their top-down
analysis was that their observed CO/CO2 enhancement ratio
of ll±2ppb CO/ppm CO2 was greater than the 8.6 ppb
CO/ppm CO2 calculated from the inventories. Wunch et al.
[2009] contrasted these top-down assessments to a bottom-up
estimate of 260 Gg CH4/yr using the statewide 2006 CARB
GHG inventory apportioned by population after removal
of agricultural and forestry emissions, and concluded that
140-340 Gg CH4/yr were not accounted for in the CARB
CH4 inventory for the SoCAB.
[4] Hsu et al. [2010] took a similar top-down approach
and used observed atmospheric enhancement ratios of
CH4 to CO from in situ whole air samples taken at Mount
Wilson (34.22°N, 118.06°W, 1770 m above sea level),
scaled by the projected CARB CO inventory for 2008, to
derive CH4 emissions of 200 ± 10 Gg CH4/yr for just the
Los Angeles (L.A.) County (Figure lb) portion of the
SoCAB (L.A. County fl SoCAB). They used methods
prescribed by the Intergovernmental Panel on Climate
Change (IPCC) to create the CARB GHG inventory and
reached a bottom-up estimate of 140 Gg CHi/vr. or 60 Gg
less than their top-down calculation for the L.A. County
portion of the SoCAB. Hsu et al. [2010] used higher spatial
resolution emissions data from CARB to construct
their bottom-up inventory and therefore did not have to
rely on population apportionment methods used by Wunch
et al. [2009],
[5] The difference between the top-down CH4 emissions
reported by Wunch et al. [2009] and by Hsu et al. [2010]
(400 Gg and 200 Gg, respectively, both based on the CARB
CO inventory) are in part due to the different geographic
areas for which they calculate CH4 emissions, and in part
due to differences in observed CH4/CO enhancements
between these two studies: 0.66 ± 0.12 mol/mol for Wunch
et al. [2009] [Wennberg et al., 2012] and 0.52 ± 0.02 mol/mol
for Hsu et al. [2010], Both works suggested that fugitive
losses of natural gas (NG) could be the source of the CH4
missing from the bottom-up inventories.
[6] More recently, Townsend-Small et al. [2012] analyzed
stable CH4 isotope ratios in atmospheric samples taken at
Mount Wilson and elsewhere in the western L.A. basin
and showed they were consistent with isotope ratios in
natural gas sources. Wennberg et al. [2012] used the
different atmospheric ethane/CH4 enhancement ratios
observed from research aircraft during the Arctic Research
of the Composition of the Troposphere from Aircraft and
Satellites (ARCTAS) field project in 2008 and the California
Research at the Nexus of Air Quality and Climate Change
(CalNex) field project [Ryerson, 2013] in 2010 to estimate
an upper limit of 400 Gg CH4/yr from natural gas leakage
in the SoCAB. Further, their top-down analysis resulted in
a calculated total emission of 440 Gg CH4/yr in the SoCAB.
Wennberg et al. [2012] also recalculated the data used by
Hsu et al. [2010] to derive CH4 emissions for the entire
SoCAB and calculated a SoCAB CH4 emission from 2008
using data from ARCTAS. The results are summarized in
Table 1.
[7] Here we use ambient measurements in the SoCAB
taken in May and June 2010 aboard the National Oceanic
and Atmospheric Administration (NOAA) P-3 research
aircraft during the CalNex field study to derive CH4
emissions from the SoCAB using methods different from
Wennberg et al. [2012], We further examine CH4 emissions
from landfills and dairy farms in the SoCAB identified in the
bottom-up CH4 inventories reported by Hsu et al. [2010]
and Wennberg et al. [2012], We then expand on these
previous studies by examining light alkane emissions from
Los Angeles area data sets. In addition to CH4 and ethane,
we examine propane, n- and i-butane, and n- and i-pentane
measurements to derive emissions of each of these light
alkanes in the SoCAB, and use them in a system of linear
equations to further quantify the source apportionment of
CH4 in the L.A. basin.
2. Measurements
[8] We use trace gas measurements from a subset of
platforms and sites from the CalNex field study. The NOAA
P-3 research aircraft flew all or parts of 16 daytime flights in
and around the L.A. basin. Two independent measurements of
CH4 and CO2 were made aboard the aircraft by wavelength-
scanned cavity ring-down spectroscopy (WS-CRDS; Picarro
1301 m) [Peischl etal., 2012], and by quantum cascade laser
direct absorption spectroscopy (QCLS) [Kort et al., 2011],
4976
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
Imprecision of the 1 Hz Picarro CH4 measurement is ±1.4
ppbv (all uncertainties herein are 1-s), and inaccuracy is
estimated at ±1.2 ppbv. Imprecision of the 1 Hz QCLS
CH4 measurement is ±1 ppbv, and inaccuracy is estimated
at ±15 ppbv. Imprecision of the 1 Hz Picarro CO2 measure-
ment is ±0.14ppmv, and inaccuracy is estimated at
±0.12ppmv. Imprecision of the 1 Hz QCLS CO2 measure-
ment is ±0.05 ppmv, and inaccuracy is estimated at
±0.10 ppmv. All CH4 and CO2 measurements are reported
as dry air mole fractions. For this work, CH4 and CO2 data
from the Picarro instrument are used, and QCLS CH4 data
from May 8 are used when the Picarro instrument was not
operating. The 1 Hz CO data used in this analysis were
measured by vacuum ultraviolet fluorescence spectroscopy
[Holloway et al., 2000], Imprecision of the 1 Hz CO data
is ±1 ppbv; inaccuracy is estimated at ±5%. C2 to C5
alkanes, and their structural isomers, were measured in
whole air samples [Colman et al., 2001], periodically filled
during flight. Imprecision of these alkane measurements
is ±5%; inaccuracies are estimated at ±10%. Wind
measurements were derived from various sensors aboard
the NOAA P-3; the uncertainty of the 1 Hz wind speed is
estimated to be ±1 m/s. Sensors aboard the NOAA P-3 also
measured relative humidity, ambient temperature, and
potential temperature with an estimated 1 Hz uncertainty
of ±0.5 °C, ±0.5 °C, and ±0.5 K, respectively.
[9] At the CalNex Pasadena ground site, located on
the California Institute of Technology (Caltech) campus,
measurements of C2-C5 alkanes were made by a gas
chromatograph-mass spectrometer on 5 min integrated
samples taken every half hour [Gilman et al., 2010],
Imprecision of these measurements are ±8% for ethane
and ±6% for propane; inaccuracy is estimated at ±15%
for each. Data from the ground site were taken between
15 May and 15 June 2010. CH4was not measured at the
Pasadena ground site.
[10] Additionally, whole-air flask samples were taken
twice daily at the Mount Wilson Observatory (MWO) for most
days during May and June 2010 and analyzed for a variety of
trace gas species, including CH4, CO2, CO, and hydrocarbons
[Dlugokencky et al., 2011; Conway et al., 2011; Novelli and
Masarie, 2010], Imprecision of the CH4 measurement is
±1 ppb; imprecision of the CO2 measurement is ±0.1 ppm;
imprecision of the CO measurement is ±1 ppbv, and inaccu-
racy of the CO measurement is estimated to be ±5%.
[11] We also analyze alkane data from whole air samples
taken in the L.A. basin prior to 2010. Ethane and propane
were measured in whole air samples taken on four flights
in L.A. aboard an instrumented National Aeronautics and
Space Administration (NASA) DC-8 research aircraft
during ARCTAS in June 2008 [Simpson et al., 2010],
Ethane and propane were also measured on one flight in
L.A. aboard the NOAA P-3 during the Intercontinental
Transport and Chemical Transformation (ITCT) study in
May 2002 [Schauffler et al., 1999],
3. Methods
[12] To ensure sampling from the L.A. basin, we consider
aircraft data collected between 33.6 and 34.3 °N latitude and
118.5 and 116.8°W longitude (Figure Id, dashed box) in the
following analysis. Aircraft data were further limited to
samples taken between 1000 and 1700 PST, between 200
and 800 m above ground, and below 1400 m above sea level,
to ensure daytime sampling was within the well-mixed
boundary layer, which averaged 1000 ± 300 m above
ground level for the daytime L.A. flights [Neuman et al.,
2012], Ground-based measurements at Pasadena were
retained between 1000 and 1700 PST to ensure sampling
of a well-mixed daytime boundary layer. For MWO
measurements, afternoon samples, which typically occurred
between 1400 and 1500 PST, were retained to capture
upslope transportation from the L.A. basin [Hsu et al.,
2010], Linear fits to the data presented below are orthogonal
distance regressions [Boggs et al., 1989] weighted by
instrument imprecision (weighted orthogonal distance
regression (ODR)). The total uncertainty in the fitted slope
is calculated by quadrature addition of the fit uncertainty
and the measurement uncertainties.
[13] For flux determinations, crosswind transects were
flown downwind of known point sources. Enhancements of
CH4 above background levels were integrated along the flight
track, and a flux was calculated using the following equation:
flux Vt n cos6at> nflzMz Xm6yMy (1)
Zo -y
where v cos(a) is the component of the average wind velocity
normal to the flight track, n is the number density of the
atmosphere, z0 is the ground level, zi is the estimated
boundary layer height, and Xm is the measured mixing ratio
enhancement above the local background along the flight
track [White et al., 1976; Trainer et al., 1995; Ryerson
et al., 1998; Nowak et al., 2012], Boundary layer heights
are estimated from vertical profiles of relative humidity,
ambient temperature, and potential temperature made prior
to and after the crosswind transects. We assume the plume
is vertically homogeneous within the mixed layer at the
point of measurement, and the wind velocity is constant
between emission and measurement. We estimate the uncer-
tainty in these assumptions, combined with the uncertainties
of the wind speed, wind direction, temperature, and integrated
atmospheric enhancements to be ±50% for the plumes studied
here [Nowak et al., 2012], Weighted averages of the fluxes are
calculated following Taylor [1997], When calculating the CH4
flux from dairies, CH4 variability immediately upwind of
the dairies is sufficiently large to complicate interpolation
from the downwind local background. To account for this,
we take the weighted ODR slope of CH4/CO immediately
upwind, multiply this ratio by the measured CO downwind
of the dairies, and integrate the plume CH4 enhancement
calculated from CO (CO x [CH4/CO]upwmd), similar to the
integrations performed by Nowak et al. [2012], This assumes
the dairies emit a negligible amount of CO.
[14] As with previously published works [Wunch et al.,
2009; Hsu et al., 2010; Wennberg et al., 2012], we estimate
total CH4 emissions in the SoCAB by multiplying enhancement
ratios of CH4 to CO and CO2 by inventory estimates of CO
and CO2 for that region:
\ (- \
c
„ ch4' MWCH4
Ech„ /* ~Y~ x xEx (2)
A ODR slope MWx
where ECh4 is the emission of CH4, X is either CO or CO2,
MW is the molecular weight, and Ex is the inventory
4977
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
Table 2. Inventories Used in Current Analysis
Emission Inventory Year Geographic Area
180 Tg C02/yr CARB GHGa 2009 SoCABc
979 Gg CO/yr CARBb 2010 SoCAB
301 Gg CHt/yr CARB GHGa 2009 SoCABc
a2009 CARB C02 and CH4 emissions (http://www.arb.ca.gov/cc/inven-
tory/data/ data.htm).
'projected 2010 CARB CO emissions (http://www.arb.ca.gov/app/
emsinv/fcemssumcat2009.php).
'statewide inventory apportioned by SoCAB population.
emission value of either CO or CO2. Although not necessarily
emitted from the same sources, we assume emissions of
CH4, CO, and CO2 are well-mixed by the time they are
sampled from the NOAA P-3.
[15] We use the following latest available inventories for
our analysis below: the 2010 CARB emissions inventory
for CO projected from the base-year 2008 inventory
(http://www.arb.ca.gov/app/emsinv/fcemssumcat2009.php)
and the 2009 CARB GHG inventory (http://www.arb.ca.gov/
cc/inventory/data/data.htm). Both inventories were accessed
in November 2012.
[16] CARB projects the total 2010 annually averaged CO
emissions in the SoCAB at 979 Gg CO/yr (Table 2). We use
the annually averaged CARB inventory that excludes
biomass burning CO emissions because no known biomass
burning events were observed in the L.A. basin during
CalNex. This estimate is 4% less than the summertime
CO inventory without biomass burning emissions, and
approximately 6% less than the annually averaged CO
inventory including biomass burning emissions used by
Wennberg et al. [2012], To estimate 2010 CH4emissions
in the SoCAB using the 2009 CARB GHG inventory, we
follow the method used by Wunch et al. [2009] and take
the total statewide emission of 1525 Gg CH4/yr, less
agricultural and forestry CH4 emissions of 898 Gg CH i/yr. then
apportion the remainder by population. In2010,the SoCAB
comprised 43% of California's population (http://www.arb.
ca.gov/app/emsinv/trends/ems_trends.php). However, unlike
Wunch et al. [2009], we include SoCAB dairy emissions
of 31.6Gg CH4/yr, which are calculated in section 4.3
below. Therefore, we attribute a total of 301 Gg CH4/yr to
the SoCAB based on the 2009 CARB GHG inventory
(Table 2).
[17] According to CARB's mobile source emission
inventory for the Los Angeles County portion of the SoCAB
(http://www.arb.ca.gOv/jpub/webapp//EMFAC2011WebApp/
emsSelectionPage l.jsp), mobile source CO2 emissions
remained essentially unchanged between 2009 and 2010
(39.94 versus 39.95 Tg C02/yr). Additionally, the statewide
CARB GHG inventory for CO2, with out-of-state electricity
generation emissions removed, decreased by less than 2% be-
tween 2008 and 2009. Therefore, we assume errors due to
sampling year are negligible in examining the CO2 emission
inventories in the SoCAB from 2009 to 2010. To estimate
2010 CO2 emissions in the SoCAB using the 2009 CARB
GHG inventory, we take the total statewide emission of
465.7 Tg C02/yr, subtract out-of-state electricity generation
of 47.9 Tg C02/yr, and then apportion the remainder by
population. We therefore attribute 180 Tg CO2AT to the
SoCAB using the 2009 CARB GHG inventory (Table 2).
We do not compare to the Vulcan CO2 inventory [Gurney
et al., 2009] because at present, it is only available for the
2002 reporting year.
4. Results and Discussion
4.1. Total Derived Emission of CH4 in L.A.
and Comparison to Inventories
[is] In this section, we use P-3 measurements of CH4,
CO, and CO2 to calculate enhancement ratios representative
of the integrated emissions from the L.A. basin. We then
use tabulated CO and CO2 emissions taken from the
CARB inventories to derive total CH4 emissions based on
enhancement ratios observed in CalNex and compare to
earlier estimates of total CH4 emissions in L.A.
[19] Figure lc shows known stationary sources of CH4 in
the L.A. area, which include landfills, dairies, wastewater
treatment facilities, and oil fields, as well as the location of
measurement sites used in this study. Dairy sources are
sized by estimated CH4 emissions from enteric fermentation,
as explained in section 4.3. Landfills are sized by CH4
emissions from the 2008 CARB GHG inventory
(L. Hunsaker, personal communication, 2011). Point sources
are sized by 2009 CARB individual facility CH4 emissions
(https://ghgreport.arb.ca.gov/eats/carb/index.cfm) but do
not stand out in the map due to their low CH4 emissions
relative to the landfills and dairies. Figure Id shows the
locations of daytime boundary-layer CH4 data from the
P-3, colored by observed mixing ratio, that were retained
for the analysis as described previously. The largest
concentrations of CH4 were typically encountered along
the mountains at the north edge of the L.A. basin, likely
driven by transport of air within the basin, as typical daytime
winds in the L.A. basin were from the west and southwest
during May and June 2010 [Washenfelder et al., 2011],
CalNex CH4 data are plotted against observed CO inFigure 2a.
Weighted ODR fits to these data resulted in derived en-
hancement ratios of 0.74 ± 0.04 and 0.68 ± 0.03 ppbv
CH4/ppbv CO from the NOAA P-3 and MWO, respectively.
We note that the same CH4/CO enhancement ratio of
0.74 ±0.03 was reported by Wennberg et al. [2012] using
the CalNex P-3 data with different selection criteria. We
include box and whisker plots in Figure 2a to show that the
weighted ODR fit to the data is insensitive to the relatively
few data points of higher CH4. The ratio calculated from
the CARB inventory (Table 2) is 0.54 ppb CHi/ppb CO
and is displayed for comparison.
[20] CalNex CH4 data are plotted against observed CO2 in
Figure 2b. The slope from a weighted ODR of P-3 data is
6.70±0.01 ppb CH4/ppm CO2 and of MWO data is
6.60±0.04ppb CH4/ppm CO2. The ratio of the CARB
inventories from Table 2 is 4.64 ppb CHi/ppm CO2 and is
displayed for comparison. In this case, because CH4 and
CO2 are measured with high precision and accuracy, the
largest uncertainties in interpreting the slope as an emissions
ratio are likely determined by the extent of mixing of
emissions from different sources within the Los Angeles
air shed. Similarly, Figure 2c shows a correlation plot of
CO against CO2. The slope from a weighted ODR of P-3
data is 9.4 ±0.5 ppb CO/ppm CO2 and of MWO data is
10.4 ± 0.5 ppb CO/ppm CO2. The ratio of the CARB inven-
tories from Table 2 is 8.5 ppb CO/ppm CO2 and is plotted
4978
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
2400
2300
> 2200
.Q
Q.
2100
¦<3-
X
o
2000
1900
P-3
MWO
CARB
inventory ratio
1800
200 400 600
P-3
O MWO
CARB
inventory ratio
CH4/C02
Figure 2. Scatter plots of CH4, CO2, and CO from all 1 s data points along flight track highlighted in
Figure 1. Dots are from the NOAA P-3, while red circles are from NOAA GMD flask samples taken at
the Mount Wilson Observatory during CalNex. Weighted ODRs (solid lines) result in slopes of
(a) 0.74 ±0.04 and 0.68 ± 0.04 ppb CH4/ppb CO; (b) 6.70 ±0.01 and 6.60 ± 0.04 ppb CH4/ppm C02;
and (c) 9.4 ±0.5 and 10.4 ±0.5 ppb CO/ppm CO2 from the NOAA P-3 and Mount Wilson Observatory,
respectively. The black dottedlinesrepresentmolar ratios ofthe CARB inventorieslistedinTable2:CH4:
CO = 0.54, CH4:C02 = 4.64x 10"3, and C0:C02 = 8.5x 10~3, where the background values used are the
sameasthosedeterminedfromthefittedslopes. Also, plotted inFigure 2a are boxes (25 tli-75th percentiles),
whiskers (10th-90th percentiles), and the median (horizontal line) for distributions of CH4 data calcu-
lated for 5 Oppbv wide bins from the NOAA P-3 CO data.
for comparison. We estimate a ±7.5% uncertainty in each of
the CARB CO and CO2 inventories, which is sufficient to
explain the difference between the CO/CO2 enhancement
ratio measured from the NOAA P-3 and the ratio calculated
from the CARB inventories. Quantitative agreement be-
tween emission ratios derived from P-3 and MWO data
(Figures 2a-2c) is likely due to the fact that the transport
within the basin was driven by the land-sea breeze, meaning
typical daytime winds in the Pasadena area near Mount
Wilson were from the southwest [Washenfelder et al.,
2011], This transport, and the highest values of CH4and
CO2 in the P-3 data that are not seen at MWO (Figures 2a
and 2b), also suggests that MWO preferentially samples
the western part of the L.A. basin [Hsu et al., 2010], We
therefore use enhancement ratios determined from
the NOAA P-3 data to derive CH4 emissions from the
entire basin.
[21] We note that the ratio of the latest CARB CO and
CO2 inventories (Table 2) are in better agreement with
ambient enhancement ratios in the CalNex data than was
the case for Wunchet al. [2009], This is likely due to either
improved CARB inventories, the present use of a basin-wide
data set to determine basin-wide emission ratios, or both.
[22] With the slopes and inventory values quantified, we
next derive a CH4 emission using equation (2). Using the
CH i/CO slope derived from the weighted ODR fit to the
2010 NOAA P-3 data and the projected 2010 CARB
annually averaged CO emission inventory in equation (2) yields
an estimated SoCAB emission of 410 ± 40 Gg CH4/yr.
The stated uncertainty is the quadrature propagation of
the measurement uncertainty, errors on the slope of the
ODR fit to P-3 data, and an estimated uncertainty in the
CARB CO inventory. We note our derived emission of
410 ±40 Gg CH i/yr is similar to that derived from the P-3
data by Wennberg et al. [2012], which was 440 ± 100 Gg
CH/vr using different selection criteria. It is further consistent
with the emission derived by Wunch et al. [2009] of
400 ± 100 Gg CH4/yr, which assumed a CARB CO inven-
tory uncertainty of 15%. We also determine CH4 emissions
using estimates of CO2 emissions in the SoCAB. P-3
measurements of the CH 1/CO2 enhancement ratio observed
during CalNex and SoCAB CO2 emissions inferred from
4979
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
34.02
34 oo
33.98
z
o
0)" 33 96
T3
S 33 94
33 92
33.90
117 90 117 85 117 80 117 75 117.70
longitude, °W
19 May 2010 (PST)
Figure 3. (a) The map from Figures lc to le shows the in-
set for Figure 3b in red. (b) Five downwind transects, sized,
and colored by CH4 mixing ratio, showing enhancements in
CH4 downwind of the Olinda Alpha landfill (green outline).
Winds were from the southwest except on 14 May, when
they were from the west and southwest, (c) Example of inte-
gration of the CH4 plume from the 19 May flight. The filled
pink area is integrated above the surrounding background
(gray line). The upwind transect on this day passed down-
wind of two power plant (Electric Generating Unit (EGU))
plumes.
the 2009 CARB GHG inventory result in a derived CH4
emission rate of 440 ± 30 Gg CH4/yr, with the stated
uncertainties determined by quadrature propagation of the
measurement uncertainty, errors on the slope of the ODR
fit to P-3 data, and an estimated uncertainty in the CARB
CO2 inventory. This value, based on the CO2 inventory, is
consistent with that derived using P-3 measurements and
the CO inventory, further supporting both our assessment
of uncertainties in the CARB CO and CO2 inventories,
and our assumption of sampling well-mixed emissions in
the SoCAB. since any outlying CH4 data do not affect the
overall emission estimates significantly.
[23] The derived 2010 top-down SoCAB CH 1 emission
of 410 and 440 Gg CH4/yr reported here using the CARB
CO or CO2 inventories, respectively, are in quantitative
agreement, in contrast to that reported for 2008 [Wunch
et al., 2009], The 2010 estimates are a factor of 1.35 to
1.45 greater than the modified population-apportioned
2009 CARB GHG inventory value of 301 Gg CH4/yr
(Table 2). A concurrent inverse modeling study by Brioude
et al. [2012] has found no statistical difference between the
total SoCAB CO emissions reported by CARB for 2010 and
a top-down approach that estimated CO emissions in the
SoCAB region using the same CO measurements used in
this paper. For this reason, and for consistency with
published works |Wunch et al., 2009; Hsu et al., 2010;
Wennberg et al., 2012], we use 410±40Gg CH4/yr from
the top-down CH4 assessment based on 2010 P-3 measured
CH i/CO enhancement ratios and the CARB CO inventory
for the remainder of our analysis.
4.2. Methane Emissions From L.A. Basin Landfills
[24] Landfills are the largest nonfossil fuel CH4 emission
source in the bottom-up inventories compiled by Hsu et al.
[2010] and by Wennberg etal. [2012], but these two studies
disagree on the magnitude of this source. Hsu et al. [2010]
estimated annual emissions from landfills totaled 90 Gg
CH4/yr from the Los Angeles County portion of the South
Coast Air Basin. Wennberg et al. [2012] reported landfill
emissions of just 86 Gg CH4/yr for the entire South Coast
Air Basin. However, that number is too low due to an error
in their gridded landfill emissions inventory (P. Wennberg.
personal communication, 2012) and is discarded in the
following analysis.
125] In the CARB GHG inventory, CH4 emissions are
calculated for individual landfills using methods prescribed
by the IPCC and summed over all landfills to estimate a
statewide total. Annual CH4 emission values for individual
landfills were obtained directly from CARB (L. Hunsaker,
personal communication, 2011) to facilitate direct comparison
to the P-3 data from CalNex. We use the P-3 data to calculate
emissions from two of the largest CH4-emitting landfills in
the statewide GHG inventory, both of which are located in
the SoCAB.
|> | The first landfill results we examine are from the
Olinda Alpha landfill (33.934°N, 117.841°W) in Brea,
Orange County, California. The NOAA P-3 flew five
daytime boundary-layer transects on five different days
downwind of this landfill (Figure 3), and a CH4 emission
flux was determined for each transect using equation
(1). The results are summarized in Table 3. For the three
transects when both the WS-CRDS and QCLS CH4
instruments were sampling ambient air, flux determinations
using these independent CH4 measurements agreed within
Table 3. Landfill Emission Fluxes Determined Aboard the NOAA
P-3 in 2010 From Downwind Plume Transects
Transect Flux Flux 2008 CARB GHG
Landfill Date (10M molecules/s) (Gg/yr) Inventory" (Gg/yr)
Olinda
8 May
1.13
9.5
Alpha
14 May
1.45
12.2
16 May
1.74
14.6
19 May
1.61
13.5
20 .funct!
2.90
24.3
Average
1.49±0.35
12.5±2.9
Puente
8 May
4.29
36.0
Hills:
19 May
3.62
30.4
20 June
4.48
37.6
Average1"
4.06 ± 1.18
34.0 ±9.9
"data from CARB (L. Hunsaker, personal communication, June 2011).
'weighted average, assuming a 50% uncertainty in the individual flux
determinations [Taylor, 1997],
4980
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
3%. In these cases, the flux was averaged and reported in
Table 3. Three nearby CH4 point sources are identified in
the 2009 CARB GHG inventory: an oil and gas field power
plant, which burns natural gas for fuel; the landfill power
plant at Olinda Alpha, which burns landfill gas for fuel;
and general stationary combustion from the landfill
operations. Inventory data suggest that these three sources
together emit between 0.0004 and 0.0015 Gg CHi/yr.
negligible amounts relative to CH4 emitted directly from
the landfill. On 19 May, the NOAA P-3 sampled plumes
from the nearby oil and gas power plant and the landfill's
power plant, both of which burn natural gas as fuel
(Figure 3c). A large spike in CO2, some CH4, and perhaps
a small amount of CO were encountered in the landfill
power plant plume. However, downwind of the landfill in
the large plume of CH4, the CO2 enhancement does not
stand out significantly above the background variability.
Therefore, our analysis of P-3 data supports the conclusion
from the inventory that landfill CH4 emissions dominate
the observed plume enhancements downwind of Olinda
Alpha landfill. Using NOAA P-3 CH4 data from all five tran-
sects, we directly calculate a weighted average CH4 emission
flux via equation (1) of (1.49 ± 0.35) x 1025 molecules/s,
equal to 12.5 ± 2.9 Gg CH4/yr assuming a constant
emission, where the weights are the 50% uncertainty of each
determination. For comparison, the CARB GHG inventory
emission estimate from the Olinda Alpha landfill is
11.0 Gg/yr for 2008, showing agreement within the errors
of the direct estimate using P-3 airborne data.
[27] The second landfill results we examine in depth are
from the Puente Hills landfill (34.020°N, 118.006°W) in
City of Industry, Los Angeles County, California. Of all
California landfills, Puente Hills is the largest emitter of
CH4 in the 2008 CARB GHG inventory. Nearby sources
of CH4 in the 2008 CARB GHG inventory include the
Puente Hills power plant (0.00045 Gg CHi/yr) and the
Savage Hills Canyon landfill (1.1 Gg CH4/yr), both of
which are small relative to the CARB GHG inventory of
39 Gg CH i/yr emission rate for Puente Hills. The NOAA
P-3 conducted three daytime boundary layer plume transects
from which we determine an average emission flux of
(4.06± 1.18)xl025 molecules/s, which extrapolates to
34.0 ± 9.9 Gg CH4/yr assuming a constant emission
(Table 3). Similar to the findings for Olinda Alpha, the
CARB GHG inventory of 39 Gg CH4/yr for the Puente Hills
landfill is in agreement within the errors of the direct
estimate using P-3 airborne data.
[28] Quantitative agreement between CH4 flux estimates
from the NOAA P-3 and the 2008 CARB GHG inventory
for these two examples supports the use of that inventory
to quantify total CH4 emissions from landfills in the South
Coast Air Basin. According to the 2008 CARB GHG inventory,
CH4 emissions from landfills totaled 117 Gg CH4/yr in the
L.A. County portion of the SoCAB, 30% higher than the
90 Gg CH i/yr for the same geographic area using the CARB
GHG inventory in 2008 reported by Hsu et al. [2010],
which we attribute to different versions of the CARB
GHG inventory.
[29] The 2008 CARB GHG inventory further predicts an
emission from landfills of 164 Gg CH4/yr for the entire
SoCAB. On the basis of the agreement with the CARB
inventory described above for the emission rates from the
two landfills quantified directly by the CalNex P-3 data
(50 Gg CHi/yr. or 30% of the inventory total for the
SoCAB), we assume the remaining CARB landfill CH4
emission estimates are accurate.
4.3. Methane Emissions From L.A. Basin Dairies
[30] Salas et al. [2008] published dairy locations in
California for the year 2005, with an estimate of dairy cow
population for each. The locations are plotted as filled
yellow circles in Figure lc, and sized by the expected CH4
emission from enteric fermentation according to the 2009
CARB GHG inventory (144 kg CH4 per cow per year).
According to Salas et al. [2008], all dairies in San
Bernardino and Riverside counties were also located in the
SoCAB, and 87% of the dairy cows in the SoCAB in 2005
were located in the Chino area (the large grouping of dairies
in Figure lc). The Chino-area dairy operations, which at one
time were distributed across the Riverside-San Bernardino
county line in satellite images, now appear to be located
mainly in San Bernardino County as the Riverside dairies
have been converted to residential neighborhoods (e.g., see
Google Earth historical imagery since 2000). This declining
number of dairies is confirmed by the United States Depart-
ment of Agriculture (USDA) (http://www.nass.usda.gov/
Statistics_by_State/California/Publications/County_Estimates/
2010051vscef.pdf), which reports a decrease in dairy cows in
San Bernardino and Riverside Counties from 200,000 head
in 2005 to 137,500 head in 2010. In addition to dairy cows,
dairies also stock immature heifers. Further, there are beef
operations in the SoCAB, but these are negligible compared
to the San Bernardino and Riverside dairy populations.
According to the USD A, there were a total of 431,000 cattle
in San Bernardino and Riverside counties in 2005, and
295,000 cattle in2010. For both years, dairy cows represented
approximately 46.5% of the cattle population in the SoCAB.
From these dairy and cattle populations, we construct a
bottom-up emissions inventory for the SoCAB using the
same emission factors as the CARB GHG inventory.
[31] We begin with CH4 emissions from enteric fermenta-
tion. We assign to each of the 137,500 dairy cows in the
SoCAB an emission factor of 144 kg CH4/yr. We assume
the remaining 157,500 head are dairy replacements, and
assign each an emission factor of 57.7 kg CH4/yr, or the
average emission factor for 0-1 and 1-2 year old dairy
replacements in the CARB GHG inventory. We calculate
a total of 28.9 Gg CH4/yr emitted solely from enteric
fermentation in the SoCAB.
[32] In addition to enteric fermentation, manure management
practices have a substantial effect on CH4 emissions from
livestock operations. In the L.A. basin, dairies typically
practice solid storage (http://www.aqmd.gov/rules/doc/
rll27Zprll27_tasklipt_20020101.pdf and http://www.arb.
ca.gov/planning/sip/sjv_report/addtl_resources.pdf), which
emits relatively low levels of CH4 (17 kg/yr per cow)
according to the 2009 CARB GHG inventory. The tradeoff
for this practice is that it emits larger amounts of NH3 than
other types of manure management (http://www.epa.gov/
ttn/chief/ap42/ch09/draft/draftanimalfeed.pdf). Therefore,
if we attribute dry manure management emissions to the
SoCAB dairy cow population, and the dry lot emission rate
of 2.1 kg CH4/yr for the remaining heifers, we get an
additional 2.7 Gg CH4/yr from dairy operation manure
4981
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
10-
8-
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
18
15
¦
2010 P-3 CalNex L.A.
¦
2010 CalNex Pasadena
0
2008 DC-8 ARCTAS L.A.
0
2005 Baker et al. (2008)
~
2002 P-3 ITCT L.A.
ethane, ppbv
- focal wells
local seeps
pipeline-quality
dry NG
~~— California traffic
LPG/propane
Figure 5. Correlation plot of propane versus ethane from
four Los Angeles data sets. Also, plotted are composition
ratios of local wells (gray lines) and local seeps (salmon lines)
reported by Jeffrey et al. [1991], the composition ratio of
pipeline-quality dry natural gas (black dashed line), the
propane/ethane emission ratio from a San Francisco Bay-area
tunnel study reported by Kirchstetter et al. [1996], and the
average composition ratio of liquefied petroleum gas
(LPG) or propane (green line).
samples from various aircraft projects in the Los Angeles re-
gion (ITCT 2002, ARCTAS 2008, and CalNex 2010), as
well as measurements from the CalNex Pasadena ground
site in 2010. Also, plotted are lines representing the
composition ratios of other possible sources of ethane and
propane in Los Angeles.
[39] The L.A. basin is home to oil and gas operations
(Figure lc); the composition ratios depicting possible
emissions from local natural gas (gray lines) and local
geologic seeps (salmon lines) in Figure 5 are those reported
by Jeffrey et al. [1991], The lower propane content relative
to ethane seen in the seeps (e.g., the La Brea tar pits)
compared to the local natural gas is attributed to near-surface
microorganisms forming shorter chain alkanes from longer
chain alkanes during the time the natural gas migrates
toward the surface [Jeffrey et al., 1991], The average
propane/ethane ratio for processed gas in SoCalGas
pipelines [Wennberg et al., 2012] is plotted as a dashed
black line. Pipeline-quality dry natural gas has a low
propane/ethane ratio because the natural gas has been
processed (i.e., the higher alkanes have been removed from
the natural gas) before distribution. The SoCalGas ratio
is representative of natural gas piped in from out of state
(e.g., from Texas, Wyoming, and Canada); approximately
90% of natural gas used in California is imported (http://
www.socalgas.com/regulatory/documents/cgr/2010_CGR.
pdf). The on-road emissions are taken from a San Francisco
Bay-area tunnel study by Kirchstetter et al. [1996], who
reported a vehicular emission ratio of 0.13mol propane/
mol ethane roughly similar to those by Fraser et al.
[1998] (0.27 mol propane/mol ethane) and by Lough et al.
[2005] (0.06-0.18 mol propane/mol ethane). Vehicle engine
exhaust typically contains small, decreasing amounts of
CH4, ethane, and propane due to incomplete combustion
as gasoline and diesel fuel do not contain significant
amounts of these light alkanes. The on-road emissions, local
geologic seeps, and the pipeline-quality dry natural gas
from SoCalGas contain three to five times more ethane than
propane and therefore cannot alone explain the ambient
ratios measured in the L.A. basin. The propane and ethane
composition of unprocessed natural gas from local wells,
on the other hand, closely matches the SoCAB ambient
measurements from three aircraft campaigns, the CalNex
ground site measurements, and the Baker et al. study
[2008], Propane and ethane were also typically enhanced
at the same time, with the exception of one sample with
elevated propane near the Long Beach area (Figure le).
[40] The data in Figure 5 suggest that local oil and gas
wells contribute significantly to the atmospheric propane
burden in the SoCAB. However, Wennberg et al. [2012]
invoked a large source of propane from fugitive losses from
the liquefied petroleum gas (LPG) industry (i.e., propane
tanks), in addition to leaks from the pipeline-quality dry
natural gas distribution system in the L.A. basin. This would
be consistent with past works that have found significant
fugitive losses of propane in other cities, such as Mexico City
[Blake and Rowland, 1995], We therefore extend our analysis
to incorporate ethane, propane, and C4 (n- and i-butane) and
Cs (n- and i-pentane) isomers to better attribute and quantify
the sources of light alkanes and CH4 to the SoCAB atmosphere.
Light alkanes are plotted in Figure 6, with lines depicting
the composition of natural gas in SoCalGas pipelines
[Wennberg et al., 2012] and of on-road emissions [Kirchstetter
et al., 1996], We neglect chemical processing of these long-
lived alkanes (t > 3 days at OH = 1 x 106 molecules/cm3) as
we find no detectable difference between daytime and
nighttime enhancement ratios relative to CO, similar to the
findings of Borbon et al. [2013] for n-butane and CO at the
CalNex Pasadena ground site. Atmospheric enhancement
ratios of propane, n-butane, and i-butane (Figures 6b-6d)
relative to ethane are consistent with emissions having
the composition of local natural gas [Jeffrey et al., 1991],
On-road emissions do not appear to contribute significantly
to the CH4, ethane, and propane in the L.A. atmosphere, and
pipeline-quality dry natural gas and/or local geologic seeps
do not appear to contribute significantly to the propane and
n-butane relative to ethane in the L.A. atmosphere. Based on
these observations, we conclude that the local natural gas
industry contributes a significant fraction to the total
atmospheric C2-C4 alkane abundances, including propane,
in the L.A. basin. We infer CH4 emissions from the
local natural gas industry are non-negligible as well, as
discussed below.
4.6. Source Attribution
[41] Here we quantify total emissions of C2-C5 alkanes in
the L.A. basin by multiplying their observed enhancement
ratios to CO by the CARB SoCAB emission inventory for
CO. Figure 7 shows C2-C5 alkanes plotted versus CO with
their respective ODR fits. The slopes from these fits are used
in equation (2) along with the projected 2010 CARB CO
4983
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
CT3
-Q
C
P-3 CalNex 2010 L.A.
• SoCalGas (pipeline-quality dry NG)
¦ Kirchstetter et al. [1996] (traffic)
Jeffrey etai [1991] local L.A. wells (local NG)
Jeffrey et al. [1991] local L.A. seeps
i 1 r
1800 1900 2000 2100 2200
CH4, ppbv
1 1 r
0 2 4 6 8 10
ethane, ppbv
7 r
2 4 6
ethane, ppbv
>
-Q
Cl
Cl
03
Q.
O
0 2 4 6 8 10
ethane, ppbv
1 r
2 4 6 8 10
ethane, ppbv
>
Q.
Q.
T
1 2 3
n-butane, ppbv
Figure 6. Plots of CHi and C2-C5 alkanes from the NOAA P-3 CalNex data set, selected for the SoCAB
(black circles). Nighttime and high-altitude data are included. Also, included for reference are the
emission ratios of mobile sources from Kirchstetter et al. [1996] (blue line), composition ratios measured
by Jeffrey et al. [1991] for local natural gas (gray lines) and local geologic seeps (salmon lines), and
composition ratios from pipeline-quality dry natural gas (NG) delivered by SoCalGas (dashed black line).
These ratios were plotted from daytime background levels.
inventory to calculate aimual alkane emissions in the
SoCAB. We assume the slopes represent a direct emission
with no chemical aging. These emissions are listed in
the rightmost column of Table 4. Also, listed in Table 4
are the estimated contributions from mobile sources in the
SoCAB, using C1-C5 to CO emission ratios from
Kirchstetter et al. [1996] (modified as discussed below)
and CO emissions from the mobile sources category in the
projected 2010 CARB CO inventory, equal to 925 Gg CO/yr,
inequation (2).
[42] Wetmberg et al. [2012] attributed the inventory CH4
shortfall [Wunch et al., 2009; Hsu et al., 2010] by ascribing
much of the CH4 and ethane enhancements to fugitive losses
of processed pipeline-quality dry natural gas. They further
suggest the majority of atmospheric propane is due to LPG
industry/propane tank fugitive losses. Here, we consider
other possible explanations of the sources of CH4 and light
alkanes in the L.A. basin for the following two reasons.
First, the source attribution by Wennberg et al. [2012] leaves
little room for CH4 emissions from landfills, wastewater
4984
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
200 400
CO, ppbv
600
l 1 r
200 400
CO, ppbv
600
Figure 7. (a-f) Daytime measurements of alkanes versus CO from the NOAA P-3 in the L.A. basin
during CalNex are plotted as filled circles. For comparison, the alkane/CO emission ratios from a
San Francisco Bay-area tunnel study [Kirchstetter et al., 1996] are plotted as a solid blue line, which
extends to the right axis. The slope from a weighted ODR (given as ppt alkane/ppb CO), total slope
uncertainty, and R2 are given in each panel.
treatment plants, and dairies in the L.A. basin. This solution
seems unlikely based on direct emissions flux estimates
using the P-3 data downwind of landfills and dairies in
the SoCAB, as described above. Second, the attribution
by Wennberg et al. [2012] would leave a shortfall in both
n- and i-butane emissions that cannot be explained by
gasoline evaporation or emissions from mobile sources. We
use a multivariate approach based on a linear combination
of the CH4 and light alkane compositions from known
sources in order to attribute and quantify total CH4 and
C2-C5 alkane emissions in the South Coast Air Basin.
[43] We include seven different source types (sectors)
with distinct and known CH4 and C2-C5 alkane compositions
(Figure 8) in the following analysis: (1) Leaks of processed
dry natural gas from pipelines, and/or emissions from
local geologic seeps (this approach cannot distinguish
between pipeline-quality dry natural gas and local seeps);
(2) CH4-dominated emissions, such as from landfills.
wastewater treatment plants, and dairies; (3) Leaks of
unprocessed, local natural gas; (4) Leaks of liquefied
petroleum gas from propane tanks; (5) On-road combustion
emissions from mobile sources; (6) Emissions of CH4 and
C2-C5 alkanes in the SoCAB from other source sectors;
and (7) Evaporative emissions from gasoline. These are
described briefly below.
[44] 1. The South Coast Air Basin contains 14.8 million
people, and SoCalGas delivers approximately 11 Tg/yr of
natural gas to the Los Angeles area. Additionally, the
Earth's natural degassing is a known source of CH4, ethane,
and propane to the atmosphere [Etiope et al., 2008; Etiope
and Ciccioli, 2009], and the L.A. basin contains abundant
geologic hydrocarbon reserves [Jeffrey et al., 1991], We
group fugitive losses from processed pipeline-quality dry
natural gas with the emissions from local geologic seeps
because the Ci-C4 emissions from these sources are not
sufficiently different to be treated separately in our linear
4985
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
Table 4. Derived Emissions in the South Coast Air Basin (In Gg/yr) for 2010 From Each Source Sector Used in Linear Analysis
Pipeline-Quality
Estimated
Dry
CEL-Dominant
Local
Evaporated
Mobile
CARB
Summed Source
SoCAB
NG/Local Seeps (Landfills, Dairies, Etc.)
NG
LPG/Propane
Gasoline
Sources
Other
Totals
Totar
ch4
192 ±54
182 ±54
32 ±7
_
_
4.9± 1.3
1.2±0.3
411 ± 77
41 lb± 37
Ethane
5.9± 1.7
-
4.5± 1.0
0.05 ±0.02
0.0±0.0
0.6±0.1
0.3±0.1
11.4 ± 1.9
11.4b± 1.6
Propane
1.5 ± 0.4
-
9.9±2.0
6.6±2.9
0.006 ±0.001
0.1±0.0
1.6±0.4
19.8±3.6
19.8±2.7
n-Butane
0.3±0.1
-
5.9± 1.2
0.02±0.01
0.5±0.1
0.3±0.1
1.4±0.4
8.5 ± 1.3
8.3± 1.2
i-Butane
0.3±0.1
-
2.2±0.5
0.13±0.06
0.08±0.02
0.04±0.01
1.8±0.5
4.6±0.6
5.1±0.7
n-Pentane
0.07 ±0.02
-
2.2±0.5
-
2.6±0.4
1.0 ± 0.1
0.3 ± 0.1
6.6±0.6
6.5±0.9
i-Pentane
0.11 ± 0.03
-
2.4 ± 0.5
0.003 ± 0.001 7.6 ± 1.0
3.9 ± 0.5
0.03 ± 0.01 14.1 ± 1.2
14.1 ± 1.8
'includes measurement, ODR fit, and inventory uncertainty.
bWennberg etal. [2012] estimate emissions to the SoCAB of440 ± lOOGg CH4/yrand 12.9±0.9Ggethane/yr.
combination analysis (illustrated by the similarity in slopes
of the dashed black and salmon-colored lines in Figure 6).
Both pipeline-quality dry natural gas and local seep
emissions contain similar amounts of CH4 and ethane
relative to one another and have less C3-C5 alkanes relative
to ethane than local, unprocessed natural gas. For pipeline-
quality dry natural gas, most C3+ alkanes are removed
during the processing stage, which is typically done close
to the source, which for -90% of the natural gas used in
California is in Canada, Wyoming, and/or Texas. For local
seeps, most C3+ alkanes are either preferentially adsorbed
in shallow sediments compared to CH4 or biodegraded
by microbes in the Earth's crust during the seepage of local
natural gas to the surface [Jeffrey et al., 1991], We use
SoCalGas samples of pipeline-quality natural gas from
2010 [Wennberg et al., 2012] to represent this source and
estimate the uncertainty of the composition at 15%.
[45] 2. CH4-dominant emission sources, which for this
analysis include landfills, wastewater treatment plants, and
livestock, emit CH4but no significant amounts of C2-C5
alkanes. This is represented in our analysis as a unit vector
containing only CH4.
[46] 3. From 2007 to 2009, the oil and gas industry in the
L.A. basin produced roughly 12-13 billion cubic feet of
natural gas per year, mostly associated gas from oil wells
(http ://www. conservation, ca. gov/dog/pub s_stats/annual_
reports/Pages/annual reports.aspx). We use an average of
the samples reported by Jeffrey et al. [1991] weighted by
2009 gross natural gas production per field and estimate
the uncertainty of this composition at 25%.
[47] 4. Two types of LPG are sold in the Los Angeles
area: One is almost completely composed of propane;
the other has traces of n- and i-butane (http://www.arb.
ca.gov/research/apr/past/98-338_l.pdf). We use the ratios
reported by Blake and Rowland [1995] from direct
analysis of LPG in Los Angeles, which is consistent
with an average of the two types of LPG sold in L.A.,
and estimate the uncertainty of the composition at 10%.
[48] 5. On-road combustion emissions are modified from
the work of Kirchstetter et al. [1996] by multiplying emission
ratios of alkanes to CO by the 925 Gg CO/yr from on-road
sources in the projected 2010 CARB CO inventory. The
C4—C5 emissions represent unburned fuel and are typically
proportional to the fuel composition; the C1-C3 emissions
typically represent incomplete combustion products. To
account for differing fuel compositions since the time of
the Kirchstetter et al. [1996] study, the i- and n-butane
emissions calculated for mobile sources in the SoCAB
(Table 4) have been scaled to the i-pentane emissions based
on their relative abundance in gasoline [Gentner et al., 2012],
[49] 6. There are additional sources of light alkanes in the
SoCAB. We use the 2010 CARB speciated inventory for
total organic gases (http://arb.ca.gov/ei/speciate/interoptlO.
htm) and projected 2010 total organic gas emissions (http://
www.arb.ca.gov/app/emsinv/fcemssumcat2009.php) for the
SoCAB to estimate emissions of light alkanes not specified
in other source sectors. These include emissions from aerosol
spray cans and other consumer products, coatings and
solvents, adhesives and sealants, and fiberglass and plastics
manufacturing. For example, propane, n-, and i-butane are
commonly used as propellants in aerosol spray cans, having
replaced CFCs in the United States in the 1970s (e.g., CARB
estimates 0.6 Gg of aerosol antiperspirant vapors were emitted
to the SoCAB in 2010, of which 0.14Gg, 0.03 Gg, and
0.15 Gg were propane, n-, and i-butane, respectively).
These emissions are summed and listed in the "CARB
Other" columninTable 4. Emissions fromnatural gas leaks,
petroleum refining, petroleum marketing (gas stations),
landfills and composting, and mobile sources are not
included in these totals, because they are accounted for
elsewhere in other source sectors. We estimate a 25%
uncertainty in the "CARB Other" inventory.
[50] 7. Emissions ratios from evaporated gasoline were
calculated from 10 gasoline samples from five Pasadena
gas stations in the summer of 2010, weighted by estimated
sales of 80% regular and 20% premium [Gentner et al.,
2012], Uncertainties are those reported by Gentner et al. [2012],
[51] First, we start with estimated annual C1-C5 emissions
in the SoCAB (rightmost column of Table 4), then subtract
modified on-road emissions [Kirchstetter et al., 1996] and
projected emissions of C1-C5 alkanes from other sources
(source sector 6, above). Next, we place the remaining
source sector characteristics into a matrix and solve for the
fraction each source contributes to the remaining alkane
observations for the L.A. basin based on each source's
relative abundances of various light alkanes. The matrix
has five columns representing the five remaining source
sectors, and seven rows containing C1-C5 alkanes. We
solve the following equation [e.g., see section 4.2 of
Kim etal., 2011]
AyXj 1/4 bi (3)
where AL| is a matrix of the C1-C5 alkane composition, i, for
the source sectors, j, defined above; Xj is the fraction each
source contributes to the total observed emissions; and bi
4986
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
500 ¦
300 ¦
100-
^ 20-
CD 15-
! io-
E 5 ¦
pipeline dry NG/
local seeps
n 1 1 1 r
CH,
landfills, dairies, etc.
LPG/propane I
ethane
propane
n-butane
/-butane
c
€
0
411
Gg/yr
11.4
Gg/yr
19.8
Gg/yr
Gg/yr
5.1
Gg/yr
6.5
Gg/yr
14.1
Gg/yr
c c c c c c c
ftr ny ny qy qy cjy ccr
£ S §¦ 5 i c c
Q>
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
the observed SoCAB alkane emission to within each
alkane's emission uncertainty.
[52] Our modeled source attribution differs from the
alkane source distribution in the L.A. basin as set forth by
Wennberg et al. [2012], From a total calculated source of
410 ± 40 Gg CH i/yr in the SoCAB, we determine that
47% comes from leaks of processed pipeline-quality dry
natural gas and/or from local geologic seeps; 44% of the
CH4 comes from the sum of landfill, wastewater treatment,
and dairy emissions; 8% from the leaks of unprocessed
natural gas from production in the western L.A. basin; and
1% from mobile sources. The attribution is presented
graphically in Figure 8. Figure 8a displays the total SoCAB
emissions as a black horizontal line in each panel, with
contributions from the different source sectors given below
the line by the filled bars. Figure 8b shows the proportion
that each source sector contributes to the derived total
emissions of each alkane.
[53] Our analysis attributes CH4 emissions of
192 ± 54 Gg CHi/yr to leaks of pipeline-quality dry processed
natural gas and/or leaks from local geologic seeps but does
not distinguish further between these two different sources.
This value is nearly a factor of 5 greater than the population-
apportioned 2009 CARB GHG emissions inventory estimate
of 40 Gg CHi/yr lost from natural gas pipelines in the SoCAB.
Our estimate of 192 Gg CH4/yr is less than the maximum
emission of 400 ± 150 Gg CH4/yr estimated by Wennberg
et al. [2012]. Our estimate would represent approximately
2% of the natural gas delivered to customers in the SoCAB
and, including storage and deliveries to customers
outside the SoCAB, 1% of the gas flowing into the basin
[Wennberg et al., 2012], These percentages would
decrease linearly with any CH4 emissions attributed to local
geologic seeps. Farrell et al. [2013] estimate up to 55 Gg
CH i/yr are emitted from the La Brea Tar Pits in western
L.A. County alone; if accurate, this would imply pipeline
leaks of only 0.7% of the gas flowing into the basin, or a
factor of at least two lower than the 2% proposed by
Wennberg et al. [2012],
[54] Our analysis attributes 182 ± 54 Gg CH4/yr in the
SoCAB to emissions from landfills, wastewater treatment,
and dairies. SoCAB landfills account for 164 Gg CH4/yr
in the 2008 CARB GHG inventory; a value supported by
our analysis in section 4.2. In section 4.3, we estimated
in a bottom-up inventory that SoCAB dairies emitted
31.6 Gg CH4/yr. Wennberg et al. [2012] estimated an
emission of 20 Gg CH i/yr from wastewater treatment. These
independent estimates sum to 216 Gg CH4/yr and are
consistent with our source apportionment using NOAA
P-3 data.
[55] CH4 emissions of 31.9 ± 6.5 Gg CH4/yr are ascribed
to leaks of local, unprocessed natural gas and would repre-
sent 17% of the local production in 2009, the latest year
for which data are available (http://www.conservation.ca.
gov/dog/pubsstats/annualreports/Pages/annualreports.aspx).
This number assumes a CH4 composition of 72.5% by volume
for natural gas produced in the South Coast Air Basin,
which is calculated as an average from the samples reported
by Jeffrey et al. [1991] weighted by 2009 production. Our
derived value of 17%, although a surprisingly high amount
of local production, is consistent with a nascent bottom-up
estimate under way at CARB. A new bottom-up inventory
survey, conducted by CARB for the calendar year 2007
but not yet incorporated into the official GHG inventory,
indicates that 109 Gg CH4/yr. since revised to 95.5 Gg CH 4/vr
(S. Detwiler, personal communication, October 2012),
were emitted throughout California by the oil and gas indus-
try via combustion, venting, and fugitive losses (Table 3-1,
http://www.aib.ca.gov/cc/oil-gas/finalreport.pdf). This updated
value is a factor of 2.5 larger than the current CARB GHG
inventory tabulation of 38 Gg CH4/yr from oil and gas
extraction for 2007 in California. CH4-specific emissions
for the South Coast Air Quality Management District in
the new CARB survey report show 24.6 Gg CH4/yr were
emitted in the SoCAB (S. Detwiler, personal communica-
tion, October 2012). According to the survey, emissions in
the SoCAB accounted for 26% of the revised statewide total
oil and gas operations CH4 emission in 2007, despite
accounting for only 4.4% of statewide natural gas production
in the basin that year (http://www.conservation.ca.gov/
dog/pubs_stats/annual_reports/Pages/annual_reports.aspx).
Thus, the survey responses suggest a CH4 leak rate of 12%
of local production in the L.A. basin. Thus, our estimate of
CH4 emissions from local natural gas for 2010 based on P-3
data from CalNex is within a factor of 1.5 of the CARB
bottom-up inventory currently in development based on
the 2007 survey. According to the survey, other oil and
gas-producing regions in California show smaller CH4 loss
rates than that from the SoCAB. For instance, statewide
losses of CH4 represent approximately 2.1% of statewide
production, and CH4 losses from the San Joaquin Air Quality
District represent approximately 1.4% of production
(from Oil and Gas Districts four and five). This indicates
that losses from natural gas production are proportionally
larger in the L.A. basin than elsewhere in the State
of California.
[56] A propane emission of 6.6±2.9Gg/yr from LPG/
propane tanks would represent approximately 1% of sales
(http ://www.aqmd. gov/ceqa/documents/2012/aqmd/finalEA/
PARI 177/1177_FEA.pdf), which is less than the ~4%
calculated by Wennberg et al. [2012], and closer to the
0.6% estimated from the document cited.
[57] Finally, our analysis suggests a resolution to the
discrepancies noted above between previous top-down
assessments and the bottom-up inventory calculations for
CH4 in the SoCAB [e.g., Wunch et al., 2009; Hsu et al.,
2010; Townsend-Small et al., 2012; Wennberg et al.,
2012], We conclude the most probable source for the excess
atmospheric CH4 is likely due to a combination of primarily
leaks, not accurately represented in the current CARB GHG
inventory, from natural gas pipelines and urban distribution
systems and/or from local geologic seeps, and secondarily
leaks of unprocessed natural gas from local oil and gas
production centered in the western L.A. basin. This finding
is based on the characteristic enhancement ratios of CH4
and the various C2-C5 alkanes consistently observed in
the L.A. atmosphere, and is further supported by the spatial
information provided by P-3 samples during CalNex.
Finally, the updated values for local oil and gas industry
emissions in the recent GHG survey commissioned by
CARB, when incorporated fully into the official CARB
GHG record, will likely help to reduce this long-standing
discrepancy between top-down assessments and bottom-
up inventories.
4988
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
5. Conclusions
[58] We use aircraft measurements of CH4, CO, and CO2
during the CalNex field campaign to show that emissions
of CH4 to the L.A. basin are greater than can be explained
by official state bottom-up inventories apportioned by
population, consistent with published work. The ratio of
the CARB CO and CO2 inventories is in better agreement
with our measurements of CO/CO2 in the Los Angeles
atmosphere than was the case for the analysis by Wunch
et al. [2009], which we attribute either to improved CARB
inventories, the present use of a basin-wide data set to
determine basin-wide emission ratios, or both.
[59] From crosswind plume transects downwind of the two
largest landfills in the basin, we determine CH4 fluxes that are
consistent with the 2008 CARB GHG inventory values, which
total 164 Gg CH i/yr emitted from all landfills in the South Coast
Air Basin. CH4 emission fluxes were also determined for
Chino-area dairies in the eastern L.A. basin Flux estimates from
these dairies ranged from 24 ± 12 to 87 ± 44 Gg CH i/yr. and
the average flux is consistent with a revised bottom-up inventory
originally compiled by Salas et al. [2008] and with previous
inventory estimates [Wennberg et al., 2012],
[60] Finally, we present a top-down assessment of C2-C5
alkane sources in the L.A. basin, and then apportion CH4
and the C2-C5 alkanes to specific source sectors in the
region. Using this source apportionment approach, we
estimate that 32 ± 7 Gg of CH4/yr, or 8% of the total CH4
enhancement observed in the SoCAB during CalNex, came
from the local oil and gas industry. This number represents
approximately 17% of the natural gas produced in the
region, within a factor of 1.5 of that calculated from a recent
survey that will be used to update the CARB bottom-up
inventory. We estimate 182±54Gg CH4/yr are emitted
by landfills, dairies, and wastewater treatment, which is
consistent with bottom-up inventories, and 192 ±54 Gg
CHi/yr are emitted of processed pipeline-quality dry natural
gas and/or from geologic seeps in the region. We further
conclude that leaks of processed pipeline-quality dry natural
gas and/or local geologic seeps, and unprocessed natural gas
from local oil and gas production are the most likely major
contributors to the previously noted discrepancy between
CH4 observations and State of California inventory values
for the South Coast Air Basin. Our findings suggest that
basin-wide mobile studies targeting CH4 and C2-C5 alkane
emissions from natural gas pipelines and urban distribution
systems, geologic seeps, and local oil and gas industry
production sites would be useful to further distinguish the
sources of CH4 in the L.A. basin.
[61] Acknowledgments. This work was supported in part by the
NOAA Health of the Atmosphere Program and by the NOAA Atmospheric
Chemistry, Carbon Cycle, and Climate Program. We thank Larry Hunsaker
of CARB for landfill CH4 emission data and Stephanie Detwiler of CARB
for the updated bottom-up oil and gas inventory data.
References
Baker, A. K., A. J. Beyersdorf, L. A. Doezema, A. Katzenstein, S. Meinardi,
I. J. Simpson, D. R. Blake, and F. S. Rowland (2008), Measurements of
nonmethane hydrocarbons in 28 United States cities, Atmos. Environ.,
42, 170-182, doi:10.1016/j.atmosenv.2007.09.007.
Blake, D. R., and F. S. Rowland (1995), Urban leakage of liquefied
petroleum gas and its impact on Mexico City air quality, Science, 269,
953-956.
Boggs, P. T., et al. (1989), Algorithm 676 - ODRPACK: Software for
weighted orthogonal distance regression, ACM Trans. Math. Software,
15,348-364.
Borbon, A., et al. (2013), Emission ratios of anthropogenic VOC in northern
mid-latitude megacities: Observations vs. emission inventories in Los
Angeles and Paris, J. Geophys. Res. Atmos., 118, 2041-2057,
doi: 10.1002/jgrd.50059.
Brioude, J., et al. (2012), Top-down estimate of surface flux in the Los
Angeles Basin using a mesoscale inverse modeling technique: Assessing
anthropogenic emissions of CO, NOx, and C02 and their impacts, Atmos.
Chem. Phys. Discuss., 12, 31439-31481, doi: 10.5194/acpd-12-31439-
2012.
Colman, J. J., A. L. Swanson, S. Meinardi, B. C. Sive, D. R. Blake, and F.
S. Rowland (2001), Description of the analysis of a wide range of volatile
organic compounds in whole air samples collected during PEM-tropics A
and B, Anal. Chem, 73(15), 3723-3731, doi:10.1021/ac010027g.
Conway, T. J, et al. (2011), Atmospheric carbon dioxide dry air mole
fractions from the NOAA ESRL carbon cycle cooperative global air
sampling network, 1968-2010, Version: 2011-10-14, Path: ftp://ftp.
cmdl .noaa. gov/ccg/co2/f lask/ event/.
Dlugokencky, E. J, et al. (2011), Atmospheric methane dry air mole
fractions from the NOAA ESRL carbon cycle cooperative global air
sampling network, 1983-2010, Version: 2011-10-14, Path: ftp://ftp.
cmdl .noaa. gov/ccg/ch4/f lask/event/.
Etiope, G, and P. Ciccioli (2009), Earth's degassing: A missing ethane and
propane source, Science, 323, doi:10.1126/science.1165904.
Etiope, G, K. R. Lassey, R. W. Klusman, and E. Boschi (2008),
Reappraisal of the fossil methane budget and related emission from geo-
logic sources, Geophys. Res. Lett, 35, L09307, doi:10.1029/
2008GL033623.
Farrell, P, D. Culling, and I. Leifer (2013), Transcontinental methane
measurements: Part 1. A mobile surface platform for source investigations,
Atmos. Environ, doi: 10.1016/j .atmosenv.2013.02.014.
Fraser, M. P, G. R. Cass, and B. R. T. Simoneit (1998), Gas-phase
and particle-phase organic compounds emitted from motor vehicle
traffic in a Los Angeles roadway tunnel, Environ. Sci. Technol,
32, 2051-2060.
Gentner, D. R, et al. (2012), Elucidating secondary organic aerosol
from diesel and gasoline vehicles through detailed characterization of
organic carbon emissions, Proc. Natl. Acad. Sci. U. S. A, 109(45),
18318-183 23, doi:10.1073/pnas,1212272109.
Oilman, J. B, et al. (2010), Surface ozone variability and halogen oxidation
throughoutthe Arctic and sub-Arctic springtime, Atmos. Chem. Phys, 10,
10,223-10,236,doi:10.5194/acp-10-10223-2010.
Gurney, K. R, D. L. Mendoza, Y. Zhou, M. L. Fischer, C. C. Miller,
S. Geethakumar, and S. de la Rue du Can (2009), High resolution fossil
fuel combustion C02 emissions fluxes for the United States, Environ.
Sci. Technol, 43, 5535-5541, doi:10.1021/es900806c.
Hollo way, J. S, R. O. Jakoubek, D. D. Parrish, C. Gerbig, A. Volz-Thomas,
S. Schmitgen, A. Fried, B. Wert, B. Henry, and J. R. Drummond (2000),
Airborne intercomparison of vacuum ultraviolet fluorescence and tunable
diode laser absorption measurements of tropospheric carbon monoxide,
J. Geophys. Res, 105(D19), 24,251-24,261, doi:10.1029/2000JD900237.
Hsu, Y.-K, T. VanCuren, S. Park, C. Jakober, J. Herner, M. FitzGibbon,
D. R. Blake, and D. D. Parrish (2010), Methane emissions inventory
verification in southern California, Atmos. Environ, 44, 1-7, doi: 10.1016/
j.atmosenv.2009.10.002.
Jeffrey, A. W. A, etal. (1991), Geochemistry of Los Angeles Basin Oil and
Gas Systems, in Active Margin Basins, Memoir 52, edited by K. T. Biddle,
pp. 197-219, Amer. Assoc. Petr. Geologists, Tulsa, Okla.
Kim, S.-W, et al. (2011), Evaluations of NOx and highly reactive VOC
emission inventories in Texas and their implications for ozone plume
simulations during the Texas Air Quality Study 2006, Atmos. Chem.
Phys, 11, 11361-11386, doi: 10.5194/acp-l 1-11361-2011.
Kirchstetter, T. W, B. C. Singer, R. A. Harley, G. R. Kendall, and W. Chan
(1996), Impact of oxygenated gasoline use on California light-duty vehicle
emissions, Environ. Sci. Technol, 30, 661-670.
Kort, E. A, P. K. Patra, K. Ishijima, B. C. Daube, R. Jimenez, J. Elkins,
D. Hurst, F. L. Moore, C. Sweeney, and S. C. Wofsy (2011), Tropospheric
distribution and variability of N20: Evidence for strong tropical emissions,
Geophys. Res. Lett, 38, L15806, doi:10.1029/2011GL047612.
Lough, G. C, J. J. Schauer, W. A. Lonneman, and M. K. Allen, (2005),
Summer and winter nonmethane hydrocarbon emissions from on-road
motor vehicles in the midwestern United States, J. Air Waste Manage.
Assoc, 55, 629-646.
Neuman, J. A, et al. (2012), Observations of ozone transport from the free
troposphere to the Los Angeles basin, J. Geophys. Res, 117, D00V09,
doi: 10.1029/2011JD016919.
Novelli, P. C, and K. A. Masarie (2010), Atmospheric carbon monoxide
dry air mole fractions from the NOAA ESRL carbon cycle cooperative
4989
-------
PEISCHL ET AL.: SOURCES OF METHANE IN L.A.
global air sampling network, 1988-2009, Version: 2011-10-14, Path:
ftp://ftp.cmdl.noaa.gov/ccg/co/flask/event/.
Nowak, J. B., J. A. Neuman, R. Bahreini, A. M. Middlebrook, J. S.
Holloway, S. A. McKeen, D. D. Parrish, T. B. Ryerson, and M. Trainer
(2012), Ammonia sources in the California South Coast Air Basin and
their impact on ammonium nitrate formation, Geophys. Res. Lett., 39,
L07804, doi: 10.1029/2012GL051197.
Peischl, J., et al. (2012), Airborne observations of methane emissions from
rice cultivation in the Sacramento Valley of California, J. Geophys. Res.,
117, D00V25, doi:10.1029/2012JDO17994.
Ryerson, T. B., et al. (1998), Emissions lifetimes and ozone formation in
power plant plumes, J. Geophys. Res., 103(D17), 22,569-22,583.
Ryerson, T. B., et al. (2013), The 2010 California Research at the Nexus of
Air Quality and Climate Change (CalNex) field study, J. Geophys. Res.
Atmos., doi:10.1002/jgrd.50331.
Salas, W. A., et al. (2008), Developing and applying process-based models
for estimating greenhouse gas and air emission from California dairies,
California Energy Commission, PIER Energy-Related Environmental Re-
search, CEC-500-2008-093, http://www.energy.ca.gov/2008publications/
CEC-500-2008-093/CEC-500-2008-093.PDF.
Schauffler, S. M, E. L. Atlas, D. R. Blake, F. Flocke, R. A. Lueb, I. M.
Lee-Taylor, V. Stroud, andW. Travnicek(1999), Distributions ofbromi-
nated organic compounds in the troposphere and lower stratosphere, I.
Geophys. Res, 104(D17), 21,513-21,535, doi:10.1029/1999ID900197.
Simpson, I. I, et al. (2010), Characterization of trace gases measured over
Alberta oil sands mining operations: 75 speciated C2-Ci0 volatile organic
compounds (VOCs), C02, CO, CH4, NO, NOy, 03 and S02, Atmos.
Chem. Phys, 10, 11,931-11,954, doi:10.5194/acp-10-11931-2010.
Taylor, I. R. (1997), An Introduction to Error Analysis, The Study of
Uncertainties in Physical Measurements, 2nd Edition, p. 174, University
Science Books, Sausalito, Calif.
Townsend-Small, A, S. C. Tyler, D. E. Pataki, X. Xu, and L. E. Christensen
(2012), Isotopic measurements of atmospheric methane in Los Angeles,
California, USA: Influence of "fugitive" fossil fuel emissions, I. Geophys.
Res, 117, D07308, doi:10.1029/2011JD016826.
Trainer, M, B. A. Ridley, M. P. Buhr, G. Kok, I. Walega, G. Htibler, D. D.
Parrish, and F. C. Fehsenfeld (1995), Regional ozone and urban plumes in
the southeastern United States: Birmingham, a case study, I. Geophys.
Res, 100(D9), 18,823-18,834.
Washenfelder, R. A, et al. (2011), The glyoxal budget and its contribution
to organic aerosol for Los Angeles, California, during CalNex 2010,
I. Geophys. Res, 116, D00V02, doi:10.1029/2011ID016314.
Wennberg, P. O, et al. (2012), On the sources of methane to the Los
Angeles atmosphere, Environ. Sci. Technol, 46(17), 9282-9289,
doi:10.1021/es301138y.
White, W. H, I. A. Anderson, D. L. Blumenthal, R. B. Husar, N. V. Gillani,
I. D. Husar, and W. E. Wilson Ir. (1976), Formation and transport of
secondary air pollutants: Ozone and aerosols in the St. Louis urban
plume, Science, 194, 187-189, doi:10.1126/science.959846.
White, A. B, C. I. Senff, A. N. Keane, L. S. Darby, I. V. Djalalova, D. C.
Ruffieux, D. E. White, B. I. Williams, and A. H. Goldstein (2006),
A wind profiler trajectory tool for air quality transport applications,
I. Geophys. Res, 111, D23S23, doi:10.1029/2006ID007475.
Wunch, D, P. O. Wennberg, G. C. Toon, G. Keppel-Aleks, and Y. G.
Yavin (2009), Emissions of greenhouse gases from a North American
megacity, Geophys. Res. Lett, 36, L15810, doi:10.1029/2009GL039825.
4990
-------
Appendix F
-------
Gas Safety Incorporated
16 Brook Lane
Southborough, Massachusetts 01772
774-922-4626 www.gassafetyusa.com
Fteport to
Damascus Citizens for Sustainability
25 Main Street
Narrowsburg, New York 12764
Fteport on a Survey of
Ground-Level Ambient Methane Levels in
the Vicinity of Wyalusing,
Bradford County, Pennsylvania
November 2013
by
Bryce F. Payne Jr.1 and Ftobert Ackley2
[This report is subject to revision.]
NOTE: Figures follow text.
There have been numerous reports of methane emissions related to shale gas
development in the vicinity of Wyalusing, Bradford County, Pennsylvania. In the
interest of furthering the understanding of those fugitive methane events
Damascus Citizens for Sustainability engaged Gas Safety, Inc. to survey
ambient air methane levels in the vicinity of Wyalusing, PA. The survey covered
parts of 9 townships on both sides of the Susquehanna River (Figure 1 -
1 Consulting and research in environmentalscience since 1992. Assoc iate
Research Professor, Dept. Env iro n m e nt a 1 Eng in e e ring and Earth Sciences,
Wilkes University, W ilke s-B a rre , PA and Senior Fellow of the Wake Forest
University Center for Energy, Environment, and Sustainability, Winston-
Salem, NC. bryce.payne@wilkes.edu
2President of Gas Safety, Inc. with 30 years experience in gas leak
detection and measurement, related regulatory compliance, and training,
bobackle y@gassafetyusa.com
-------
following text) from Towanda on the northwest to Wyalusing on the central
eastern side. Survey coverage was restricted to readily identifiable public
roadways. Consequently, the survey was most intense from the Susquehanna
River west to Pennsylvania Route 1 87.
Though the survey results do not prove a relationship between ambient air
methane contamination and groundwater contamination, it is clearly
suggestive. Further, it also suggests shale gas well operations in that area still
did not have control of the gas that has been developed there. In fact, as will
be discussed, survey data indicates there may be gas control problems in about
1 0% of the survey area resulting in elevated methane levels in most of the area.
In addition, detection of any level of methane above normal background for an
area indicates only two possible conditions: diffuse, non-point emissions are
occurring over some portion of the area, or, one or more point sources are
active within the area.
Conditions during the Survey
The survey effort involved two separate survey field work efforts, one on 31
January and the other 3-4 June 2013. Weather conditions at the time of the
January survey were not ideal. Winds were from the west at speeds consistently
near 20 miles per hour (29 feet per second). Under these conditions methane
emissions from any source disperse rapidly. Consequently, elevated methane
levels due to such emissions are more difficult to detect than under more
favorable wind conditions. Functionally this means that, during a road survey,
detection of elevated methane levels requires the sources be larger or more
intense and in closer proximity to the survey vehicle path than under more
favorable wind conditions. However, such wind conditions do cause methane
emissions to be swept along the ground surface farther and faster.
Consequently, methane emissions appear as a general elevation of methane
levels over a wider area, instead of localized markedly elevated peaks.
During the 3-4 June field work weather conditions were more favorable. The
wind was from the north-northwest at an average speed of 5 miles per hour
(around 8 feet per second). Under these conditions methane emissions would
be expected to be detectable as low concentration plumes extending for an
appreciable distance to the south-southeast of the source. Mixing layer
structure and height was not estimated during the survey, but conditions
should have favored typical lower atmospheric mixing patterns in which most
methane emissions diffuse rapidly upward.
Results of the January Survey
2
-------
As anticipated due to the wind conditions the methane levels were moderately
elevated widely over the survey area. Typical methane level observed during
the survey was low. The average methane level was 1.86 ppm, with a minimum
of 1.79 ppm, 90% were below 1.91 ppm, and 99% below 2.08 ppm.3 Under
such high wind conditions, the layer of the atmosphere that normally forms
next to the land surface4 is swept away by air that would normally move at
altitudes of a few hundred to a few thousand feet above. Under gentler wind
conditions gases released into the air tend to accumulate in plumes as they
dissipate into the turbulent but lower-wind-speed layer of air next to the land
surface. Under sustained high wind conditions the air from the higher layer
sweeps down and across the land surface rapidly sweeping any released gases
across the land surface and up into the atmosphere.
Figure 2 shows an oblique westward view of the survey area in which the data
was processed to remove values lower than 2.2 ppm and vertically exaggerate
those over 2.2 ppm by a factor of 1 000. In effect, this approach visually
defines methane levels above 2.2 ppm as elevated methane levels (EMLs). This
graphical rendering shows around 1 8 locations with elevations above 2.2 ppm.
There also appear to be many locations with EMLs near 2.2 ppm. This,
however, is an artifact of the low resolution of this image and the high
resolution of the survey data set. When this image is examined at higher
resolution most of the apparent near-2.2-ppm EMLs disappear.
To allow examination of smaller EMLs another image of data was prepared with
the methane data processed to remove values below 1.9 ppm and vertically
exaggerate values >1.9 ppm by a factor of 100. The lower 1,9-ppm cutoff and
vertical exaggeration preserved EMLs that were not apparent upon high
resolution examination of Figure 2, as illustrated by Figures 3 and 4. The
>1,9-ppm image is not shown as it is visually nearly flat at the resolution that
can be rendered on a single page of this report. In the >1,9-ppm image 57
EMLs were indentified as sufficiently clear to merit further examination (see
Appendix B for a listing of those EMLs by location). Of those 57 EMLs, 43 were
in proximity to and nearly-downwind of gas pipelines, gas well pads, farms,
industrial facilities with apparent waste water treatment ponds or lagoons.
3 During survey runs the vehicle has to make stops. The CRDS methane
instrument collects data continuously. Consequently, geographically
disproportionate amounts of data accumulate whenever the vehicle stops.
Geographically disproportionate data accumulations are removed from the data
set before statistical analysis. Images are generated using the full raw data
sets.
4 Planetary boundary layer or mixing layer. See Manhattan extended report for
more detailed discussion.NEED LINK HERE
3
-------
Further identification of the methane sources causing the other 1 4 EMLs was
beyond the scope of the survey work.
Despite the strong wind conditions a relatively large methane plume was
detected. The plume was detected over an area running from Wysox 2.5 miles
southward along the river and up to B.6 miles to the east. The plume was not
present on a later pass through the same area. The extent and consistency of
this plume over such a large area under such windy conditions, and its
relatively sudden disappearance suggest a sizeable release of methane upwind
of the plume area that ended sometime during the survey. Identification of a
likely source was beyond the scope of the survey work. It is noteworthy that
this plume was again present during the June survey. The plume may have
been related to a number of gas wells generally north of Wysox.
Conclusions from 31 January Survey
The strong wind conditions during the methane survey caused rapid mixing and
lateral dispersal of methane from any sources in or near the survey area. Under
such conditions detection of elevated methane levels is limited to those
resulting from larger emissions or those from sources in close proximity to the
roadway. The rapid mixing and lateral dispersal causes methane levels in the
area to appear more uniformly elevated than would be the case under less
windy conditions. This was indicated by the slightly elevated mean (1.86 ppm)
and narrow range of methane levels (1.79-1.91 ppm) that accounted for the
90% of the data (further discussed in comparison to the June data follows
below). All the other 1 0% of the data indicating methane levels above 1.91 ppm
occurred at less than 60 locations. Among those locations, 43 were in the
vicinity of candidate potential methane sources, in most cases gas pipelines or
gas well pads. At 14 locations with elevated methane levels candidate potential
methane sources were not readily apparent.
Results of the 3-4 June Survey
As expected under the more favorable wind conditions on 3-4 June, methane
plumes were detectable over much larger areas than during the extreme wind
conditions of the 31 January survey. Elevated methane levels occurred over
much of the survey area. Additionally the methane instrument (cavity ring
down spectrometer5) was run during travel from the survey area and during a
brief observational trip to the Leroy Township area. Those two legs of the
5 http://www.picarro.com/technology/cavity_ring_down_spectroscopy
4
-------
survey trip provided methane measurements in geographically and geologically
adjacent areas that can be reasonably regarded as comparable areas with
limited or no shale gas well activity. That area is referred to as the Reference
Area in the remainder of this report. It includes data from valleys, along a river,
and two town/city areas. Hence, the Reference Area can be reasonably
considered to have all likely natural and human-caused methane sources
typical for the geographical/geological area, but with minimal large-scale
agricultural, industrial or shale gas sources. Also, of some interest is
recognition that the methane survey work included parts of two areas under
Pennsylvania Department of Environmental Protection Consent Orders. An
image displaying the results of the June survey is provided in Figure 5.
It should be borne in mind that the survey work was limited to publicly
accessible roads. The survey, therefore, measures the impacts of methane
emissions sources at considerable distances from those sources.
Consequently, seemingly minor changes, in the tenths or hundredths of a part
per million, in ambient air methane levels are of considerable importance in
locating methane emissions sources and assessing their broader area impacts.
The June survey average methane level was 1.83 ppm, with a minimum of 1.75
ppm, 90% were below 1.88 ppm, and 99% below 2.05 ppm.3 Given the
difference in wind conditions, these levels were quite similar to those seen in
the January survey. For comparison, in the Reference Area the average methane
level was 1.78 ppm, with a minimum of 1.76 ppm, 90% were below 1.79 ppm,
and 99% below 1.81 ppm.3 Since much of the survey area is affected by the
same type and frequency of methane sources that occur in the Reference Area,
one would expect that much of the survey area data would be similar. This
was, in fact, found to be the case. It can be seen in Figure 6 that in the
Reference Area 97% of the methane levels were below 1.8 ppm, while in the
survey area in June, 37% were, but in the survey area in January less than 1 %
were below 1.8 ppm. These results suggest that methane emissions in about
37% of the survey area are effectively similar to the Reference Area. The strong
winds during the January compared to the June survey were probably the cause
of the apparent reduction in total area with readings below 1.8 ppm (37% of the
area in June compared to <1 % in January), Emissions that on 3-4 June were
rising into the air more normally, whereas on 31 January emissions were being
rapidly mixed and swept over the land surface by the strong winds.
Looking at another methane value of interest, the maximum methane level
measured in the Reference Area was 1.88 ppm. In the survey area on 3-4 June
1 0% of the measurements exceeded the Reference Area maximum, and on 31
January 16%. Consequently, it is reasonable to conclude that at least 1 0% of the
survey area is impacted by methane sources that do not occur in the Reference
Area. As previously mentioned, these are agricultural and industrial sources.
Field observations and examination of satellite imagery allowed determination
5
-------
that some of the methane sources causing the elevated methane were
agricultural or industrial, other than shale gas development. The plumes of the
ag/industrial sources appeared less extensive than the plumes of the sources
associated with shale gas development. Most of the shale gas methane
emissions sources appeared likely to be well pads and pipelines.
With regard to the relationship between ambient air methane surveys and
locations of methane sources potentially impacting an area, it is interesting to
consider the survey covered parts of the areas under two PaDEP Consent
Orders. Those two Orders were between the PaDEP and Chesapeake
Appalachia, LLC, dated 16 May 2011 6. The two Orders were designated for
impact areas referred to by PaDEP as Paradise Road and Sugar Run. It should
be borne in mind that at the time of the survey, the Consent Order impact areas
were not specifically known to GSI and were not specifically targeted. The
general outline of the survey area was selected by DCS based on reports in the
media and from residents. The specific area was determined by the operational
conditions GSI encountered in the field. Consequently, the survey covered the
Consent Orders impact areas only coincidentally. Still the survey did include
about 2/3 of the Paradise Road and Vz of the Sugar Run Consent Order impact
areas. It can be readily observed in Figure 5 that elevated methane levels were
concentrated within the Paradise Road impact area compared to the remainder
of the survey. There were elevated methane levels in other parts of the survey
area but the concentration in the central part of the Paradise Road impact area
is distinct. Though this does not prove a relationship between ambient air
methane contamination and groundwater contamination, it is clearly suggestive.
Further, it also suggests shale gas well operations in that area still did not have
control of the gas that has been developed there. In fact, as already mentioned,
the survey data indicates there may be gas control problems in about 10% of
the survey area resulting in elevated methane levels over 60-90% of the area.
In addition, detection of any level of methane above normal background for an
area indicates only two possible conditions: diffuse, non-point emissions are
occurring over some portion of the area, or, one or more point sources are
active within the area. Non-point sources are difficult to assess, precisely
because they are diffuse. As mentioned previously, at the end of the survey
work reported here a cursory evaluation run was made to the area of a
previously documented shale gas well impact in Leroy Township. NEED LINK
HERE That site is of interest in this discussion because on the land surface
methane emissions occur as a non-point source, with gas emerging from many
points over a area of uncertain extent. During the earlier evaluation of that site
6 This PA DEP Consent Order available HERE: https://www.dropbox.eom/s/3r34e3ggb88qxbo/
161 %20Consent%20Agreem%20Susquehana%20River.pdf
6
-------
nearly pure natural gas was encountered within inches of the soil surface, but
on the nearest road, about 100 yards away, and downwind at the time, only a
few ppm of methane were detected. Despite gas well remediation measures,
the 4 June run along the same roads confirmed methane levels remain in the
range of a few ppm, suggesting the methane migration problem still exists. A
cursory water sample test also indicated water in the area still has very high
methane levels. Methane contamination was prevalent in the area during the
prior evaluation. The Leroy Township situation is troubling with regard to
health and safety, and discouraging with regard to the capability of industry to
effectively correct gas well problems when they occur.
Point sources of methane present a slightly different set of concerns. A
substantial amount of methane is necessary to raise methane levels even
slightly over an extensive area, as measured from our survey over public roads.
If that amount of methane is being emitted at one or a few point sources, then
the concentration of methane in the vicinity of those sources will likely be
hazardous with respect to explosion or asphyxiation. Consequently, the
methane levels measured during the survey indicate there likely are point
sources associated with some shale gas wells in the area that do give rise to
hazardous conditions. Those point sources need not necessarily be at the gas
well itself, as the gas may find underground pathways to emerge in water wells,
homes or other structures, as occurred in Leroy Township, and the Paradise
Road and Sugar Run impact areas.
Conclusions
Methane from any source rapidly diffuses and rises in the air. Consequently,
detection of possible methane sources from any distance away requires
extremely sensitive measurement capabilities. The GSI survey approach takes
advantage of extremely sensitive measurement instrumentation to detect small
increases in ambient air methane levels as an indication of probable methane
emissions sources in a given area. Based on the data collected using that
equipment, we conclude that the Towanda-Wyalusing area is probably
substantially impacted by methane emissions from shale gas wells both within
and beyond the survey area, depending on wind conditions. The coincidence of
two DEP methane migration impact areas, Paradise Road and Sugar Road, and
the most marked ambient air methane levels suggests there are still gas control
problems associated with the shale gas wells there, as well as in another
documented impact area in Leroy Township also cursorily measured following
the main survey. A rapid water test in the Leroy area confirmed the water in
that area is still contaminated with methane. These survey results suggest
methane contamination continues and measures taken by gas well operators
with regard to methane migration problems that have occurred in these three
areas have likely been only partially effective.
7
-------
Figure 1. Overhead image of roads traveled during the survey of ambient air
methane levels in the vicinity of Wyalusing, PA on 31 January 201 3 (Google
Earth).
rr
jA Towanda
' v, , m
Monroe
Wyalusing
'* '^tT > % 'Pr-
image PA'wpartment of Conservation jncrNaturalResbiJrfei-PAM AP/USGfc ,UlV?k'
t^©2SRj^w&m•'*»' ^• '
Imagery Date: 9/11/2M2 41 *>7fe.JS1332* elev 1J7S ft Eye alt 12.41 mi Of
-------
On
Figure 2. View from east of 31 January 2013 survey of ambient air methane
levels in the vicinity of Wyalusing, PA. Data was processed to remove methane
data values below 2.2 ppm and multiply remainder by 1000 to enhance visibility.
-------
( 20 U Google
Image PA^Ic
| | | | Image C 201J TerraMetncs
Imagery Date: 9/11/2042 4I*>33S*?' ten -76.30444S* elev 1003 ft Eye aft 1614 ft O
Figure 3. An elevated methane level as rendered by processing of the
Wyalusing 31 January 201 3 methane survey data to remove values <2.2ppm
and multiply remainder by 1 000. Compare to same elevated methane location
in Figure 4.
10
-------
C 20H Google
Image PAlDffcftrtment of Conservation and Natural Reiources-PAMAPi'USGt,lX
. |r Image C 201) TerraMelric*
Imagery Date: 9/11/2042 4ffE>33SW Ion -76.304445' elev 1003 ft Eye alt 1614 ft O
Figure 4. An elevated methane level as rendered by processing of the
Wyalusing 31 January 201 3 methane survey data to remove values <1,9ppm
and multiply remainder by 1 00. Compare to same elevated methane location
Figure 3.
-------
Figure 5. The 3-4 June 2013 Towanda-Wyalusing Ambient Air Methane Survey.
Relative methane levels indicated in red (highest peak in image = 3.9 ppm).
Blue and orange markers indicate the Paradise Road and Sugar Run methane migration
impact areas (4-mile radius) desigated in 16 May 2011 PaDEP Consent Order.
o
Oj!» V '4.*011 • 1994
Irrupt *1 0«p4nm#ni c-t ( SjucjI - f AHA* OtC\
J • >oi?iGoc^w air
lm*^ Undut
l«t 4IM40M Ion ,'ivwnr •!»> UMII
.Google
lr» »ti 17174 h o
-------
Figure 6. Ambient Air Methane Surveys
T owanda-Wyalusing Area, PA January and
June 2013
-+-Refere ce Area (4Ju e)
-tt-31 January
-3-4J u ne
100.00
v
m
1:
E
hi
., 70.00
60.00
¦ o
50.00
o 40.00
1>.
•)
?, 30.00
E 20.00
II
u
II
> 10.00
0.00
18 1B5 19 2 2.05
X= Ambient Air Methane Concentration {ppm)
/
m
/
/
13
-------
Appendix G
-------
Gas Safety Incorporated
16 Brook Lane
Southborough, Massachusetts 01772
774-922-4626
www.gassafetyusa.com
Report to the Clean Air Council
on
8 June 2012 Field Inspection and Methane Sampling Survey
of
Parts of Leroy, Granville and Franklin Townships
Bradford County, Pennsylvania
NOTE: Specific location identification information is avoided in this report pending approval
of involved or potentially involved parties.
SUMMARY
A portable laser-based methane measurement system was used to survey methane levels in
northeastern Leroy Township, Bradford County, Pennsylvania and adjacent parts of Granville
and Franklin Townships on 8 June 2012. The methane system reports methane levels in air to
the nearest part per billion (ppb) every 3-4 seconds. During the survey over 7,600 methane
measurements were made. The survey data indicated one or more substantial methane
emissions were occurring in an area near and to the west of H Rockwell Road and Route 414
giving rise to a ground level plume that expanded to cover at least 4.2 square kilometers over a
period of 3.5 hours. The size and rate of expansion of the plume suggested large amounts of
methane were being emitted to the atmosphere. Heavily methane contaminated residential
water wells occurred in and around the same area, and documentation indicated heavy
contamination has existed for at least several months. Bubbling gas in Towanda Creek suggest
fugitive gas from shale gas wells may be travelling through faults and fractures, which also
carry local ground water, hence, impact local water wells. Collectively the data and
observations suggest natural gas has pervaded an extensive subsurface area beyond the area
where elevated ground-level methane was found during this survey effort. If that is correct,
then more surface emissions are likely and should be expected. The issues and concerns
presented in this report require more thorough investigation for confirmation and
quantification.
-------
BACKGROUND
A number of dramatic fugitive methane emissions were reported to have begun in Leroy
Township on 19 May 2012. Reports suggested a substantive loss of control of natural gas
flows from one or more of the shale gas wells in the Township may have occurred. In the
interest of verifying and developing independent documentation of the reportedly large
increases in natural gas emissions, the Clean Air Council ("CAC", Philadelphia, PA) contracted
Gas Safety, Inc. ("GSI", Southboro, MA) to do a one-day sampling and area visit to ascertain the
locations of observed or suspected natural gas emissions. The intention was for GSI to use a
customized, portable Cavity Ring-Down Spectrometry (CRDS) methane measurement
instrument to investigate and document the occurrence (or not) of the reported emissions.
A major concern was to perform the assessment as soon as possible to better ascertain the
possible initial intensity and extent of the event; that is, this would preferably be a short notice,
rapid response effort. Other concerns were GSI instrument availability and efficient use of
field time with the instrument. In order to assure a one-day effort would be as productive as
practical, GSI and CAC contacted various parties in pursuit of information regarding specific
locations of reported point-source gas emissions. Like the survey trip itself, such requests for
information had to be short notice, rapid response efforts. In the interest of openness and
sharing of information GSI proposed to provide through CAC its findings to cooperating
parties. GSI contacted the Emergency Management Agency of Bradford County, which
preliminarily offered to share its own records regarding the initial reports of the sudden onset
gas emissions event. CAC and GSI also contacted various private parties with similar data
sharing offers and rapid response requests for information and property access. Within 24
hours of such requests, and despite initially positive responses, only 3 private parties agreed to
provide information or access to suspected emissions or impacted areas on private property.
Ultimately no specific identification of or authorization for access to the actual point locations
of ongoing natural gas emissions was obtained in time for the survey. Hence, work was limited
to surveying methane levels on public roadways and verifying methane in well water in three
residences and collecting anecdotal reports on three others.
Weather conditions were mild and favorable. Barometric pressure was steady. Winds were
from the west-northwest increasing steadily throughout the day from nearly calm to a few
miles per hour by the end of the survey work.
The group involved in the methane sampling survey (listed just below) met at the junction of
Routes 414 and 514 in the northwest corner of Franklin Township at approximately 09:15 AM
on 8 June 2012.
Carolyn Knapp, Bradford County resident
Dan Natt, Bradford County resident
Matt Walker, Clean Air Council
Ron Kanter, videographer, Clean Air Council
Dr. Brian Redmond, PG, Wilkes University,
Dept. of Environmental Engineering and Earth Sciences
Bob Ackley, Gas Safety, Inc.
Dr. Bryce F. Payne Jr., Gas Safety, Inc.
-------
METHANE IN RESIDENTIAL WATER WELLS
A total of four residences (referred to as housel, house2,...) were visited, all served by on-site
wells with an interview at house 5 regarding houses 5 and 6. Housel was vacant. It was
reported that the residents had vacated due to the inconvenience of and health concerns
related to elevated levels of methane and contamination in well water. The house could not be
entered, and due to lack of power, no well water could be sampled. Concentrations of methane
in the air on the property were normal (normal background ambient air methane 1.75-1.95
ppm).
House2 was located on a farm near a gas well pad (Morse 3H and 5H wells). Most of the area
had elevated ambient air methane levels (max. 2.5 ppm) that appeared possibly associated
with animal manure accumulations on the farm. However, in areas where there was
substantial manure, the presumed likely source of the methane, ambient methane levels were
rarely above background and never exceeded 2.2 ppm. Upslope from the farm building area
ambient air methane levels were normal. At 200 meters east of the gas well pad methane
levels were normal. Down slope, toward Towanda Creek methane levels were elevated, with
three locations showing >100 ppm in the ambient air just above the surface of the creek bank.
In these instances the methane could have been biogenic or fugitive thermogenic, but sampling
conditions did not permit collection of samples for isotope analysis.
The water at the kitchen tap in house3 and house4 was supersaturated with methane. Upon
flowing from the faucet the water appeared "milky" due to the large amount of fine methane
bubbles present. The fine bubbles coalesced over a period of several seconds causing a
pronounced effervescence. High levels of methane in the gas evolved from the tap water were
confirmed with the CRDS instrument. No attempt was made to verify initial methane
concentration in the tap water, but reports of analyses of samples previously collected by
PaDEP or contractors indicated that the well water in these homes had been confirmed to
contain methane levels from 50 to 100 milligrams per liter, much greater than saturation
under atmospheric pressure (about 28 milligrams per liter). Ongoing supersaturation of well
water can only occur if there is substantial water "head" pressure in the well and the methane
is under sufficient pressure to reach aquifers under such pressure. It should also be noted that
such methane levels are sufficient to pose asphyxiation hazards if used for showering or other
high water uses in close quarters, symptoms of which the residents of these properties
reported.
An additional inquiry was made at another residence (designated house 5). This was a no-
notice contact initially to request information on ownership of the adjacent property. The
occupant at house5 reported the well water at that house was similarly heavily contaminated,
as well as the well of a close relative who lived in another nearby house (house6).
The wellheads at houses 3-6 had been equipped with passive or wind turbine vents, reportedly
by either PaDEP or gas company contractors. Such vents are not designed to prevent or treat
contamination of water in wells with conditions and methane exposures of the type that can
-------
cause such super-saturation with methane. Presumably the passive vents were installed to
prevent pressure driven flow of methane into the homes through possible underground
pathways. Though clearly better than the risk of not venting, the application of only passive
vents leave the residents under continuing risk of exposures to asphyxiating concentrations of
methane, ignore the at least substantial nuisance of having to use methane-super-saturated
water, and the potential for serious eruptive releases of methane up through the water well. In
addition, such levels of methane contamination necessarily imply the possibility of indirect
effects on water quality due to induced biological and chemical changes in the ground water
and the mineral medium through which it flows. Such effects might take months or years to
become fully apparent, and present a serious concern with regard to long-term degradation of
aquifers in areas where even less intensive methane contamination occurs.
In summary, of 6 houses visited or about which information was obtained, 5 had well water
that was supersaturated with methane. All 5 of those in which methane contamination was
observed or reported lie north of Towanda Creek. Four of the five contaminated residences
were occupied at the time of this inquiry, and at all 4, passive vents had been installed with the
foreseeable lack of effect on methane contamination of the water. The intensity of the methane
contamination seems to require more definitive treatment measures as well as efforts to
identify the source or sources of the contamination and actions to prevent long-term
degradation of aquifers.
SURVEY OF THE AREA FOR METHANE IN THE AIR
Cavity Ring-Down Spectrometry and Baseline Ground-Level Methane Data
The CRDS instrument is extremely sensitive, runs continuously, and is robust. Consequently
the unit quickly generates large volumes of highly reliable methane measurements on a
continuous basis. During the one-day area survey reported here, the instrument generated
7,697 methane measurements. In combination with similar quantities of data from prior
surveys in the eastern Marcellus Shale region, GSI has determined that a reliable (99.99%
confidence level) upper bound for background methane levels in ground level air is 1.95 parts
per million (ppm). GSI also has identified thousands of gas leaks in commercial pipelines in a
variety of settings and based on that experience has concluded that CRDS measured levels of
methane in excess of 2.05 ppm reliably indicate a natural gas leak in the surrounding area.
Based on these findings, GSI interprets methane levels above 1.95 ppm as presumptive, and
above 2.05 ppm as highly probable methane contamination. There is potential for some
biogenic sources to generate enough methane to cause such readings, but such potential
biogenic sources are usually readily identifiable, and limited in both extent and intensity in
comparison to fugitive natural gas from wells or infrastructure. When more definitive
evidence is needed, gas samples are collected and analyzed for isotopic composition for
comparison to similar data for suspected sources of contaminating gas.
The areas in Leroy, Granville, and Franklin Townships surveyed and reported here had
background levels and variations typical for the region, the lowest methane reading being
1.674 ppm (nominal accuracy of the CRDS is 0.001 ppm). Some areas of elevated methane in
-------
the air occurred near areas on farms with long-term animal manure loads. No elevated
methane levels were found for carcass handling, and other agricultural areas that might be
conventionally considered suspect for biogenic methane production. Interestingly no elevated
methane levels were measured at the nearest access (200 meters) to the natural gas well pad
(Morse 3H and 5H wells) within the area covered by this survey. This would seem a
reasonable finding given the well is new, with limited and new infrastructure, and there was
no wind during sampling in that area. With no wind and the low density of methane (half that
of air), any gas leaks comprised primarily of methane would likely rise directly upward and go
undetected without adequately close access to the vicinity of the leak.
Elevated methane levels, however, were detected as soon as the instrument was activated at
the junction of Routes 414 and 514. All of the initial 157 readings were above 1.95 ppm, 152
were above 2.00 ppm. Such sustained levels above 1.95 indicate a fugitive methane source
upwind. An initial drive and walk survey along and near Route 514 covering approximately 2
kilometers to the north and back indicated no methane above reasonable background levels.
The initially observed elevated readings at the junction of Routes 414 and 514 had diminished
when the instrument was returned to the location just over one hour later.
A driving survey west on Rt 414 (0.6 kilometers), south on Cross Road (0.5 kilometers), and
west on South Side Road (2.2 kilometers) again revealed no elevated methane levels, as did a
walking survey upslope from South Side Road, downwind from the Morse gas well pad.
Upon descending to the banks of Towanda Creek, methane levels rose above baseline in the
vicinity of the creek banks. Random sampling at three locations showed maximum methane
levels immediately above the soil surface of 133, 391, and 713 ppm. At the time of the
observations there was no basis for inferring whether the methane was more likely biogenic or
fugitive thermogenic gas. Methane levels were slightly elevated over most of the surveyed
area along the creek.
-------
The next leg of the survey involved a return east on South Side Road, then north across the
bridge, and west along Rt 414 (2.2 kilometers) and north on H Rockwell Road (1 kilometer)
(below red methane spike in image above). Methane levels were normal (indicated by
methane level markers in image above) until reaching H Rockwell Road, where slightly
elevated levels were again encountered (average of 32 readings = 2.068, range = 1.967 to
2.184ppm) northbound along the first approximately 500 meters of that road. About an hour
later, on the return trip south on H Rockwell Road and east on Rt 414 the methane levels (red
methane level markers in image above) had risen substantially and the affected area expanded
south and east. Methane levels began to rise relatively suddenly about SOOmeters north of Rt
414 from 2.01 ppm to a maximum of 21.979 ppm, then settled into a range of 10 to 14 ppm.
The area of elevated methane levels had expanded to the south and east as indicated by
measurements along Rt 414 showing levels descending from 4.620 ppm at H Rockwell Road to
2.049 ppm approximately 1 kilometer to the east. Another survey pass was made through the
area approximately 1 hour 50 minutes later driving eastbound on Rt 414 ( methane
level markers in image above). The elevated methane levels were then found to have
expanded to cover an area from Rockwell Road east along Rt 414 for 2.8 kilometers then north
along Rt 514 (2.8 kilometers) at an overall average concentration of 3.8 ppm. The data clearly
indicated that one or more methane emissions were present and releasing substantial
amounts of methane into the atmosphere probably within 500 meters to the north of Rt 414,
near and to the west of Rockwell Road along with other possible emissions occurring or
developing within the area enclosed by Rockwell Road and Rts 414 and 514. The measured
-------
plume covered an area of approximately 4.2 square kilometers, however, methane data and
wind direction indicate the plume probably extended considerably farther to the south and
east. Time was insufficient to measure the full extent of the plume to the south and east.
Gas was reported to have been bubbling up in Towanda Creek beneath the Cross Road bridge.
The bridge was visited to view the gas bubbling, if present. Upon arrival the bubbling proved
to be relatively easily observed. Batches of bubbles were rising to the surface at consistent
time intervals and locations, fairly regularly spaced along a line running roughly east-
northeast for the entire distance visible from the bridge, about 100 meters west to a somewhat
shorter distance east. The directional orientation of the line of bubbles and regular spacing
between bubbling points suggested association with a local fault or related subsurface
structure. The total volume of bubbles per batch was very roughly estimated to be at least 300
cubic centimeters. Over the visible length of the bubble line the bubbling was nearly always
occurring at one or more of the locations. Hence, the observed bubbling area was estimated to
have been releasing at least 300 cubic centimeters per second, or 18 liters per minute, or 38
cubic feet per hour.
The volume and spatial distribution of the bubbling locations make other potential
explanations, e.g., a biogenic methane source in the creek bottom, seem implausible. When the
direction of the bubbling line under the bridge was extended to the west-southwest, it
intersected the area where methane had been measured in the creek bank soils earlier in the
day, suggesting the possibility that methane emissions may have been occurring along a fault
line, but due to lack of access and time there was no opportunity to evaluate this possibility.
It is important and useful to note that the gas released in the creek under the bridge could not
be confirmed to be methane with the CRDS instrument due to wind conditions and no access to
the bubbling points in the creek due to the high elevation of the deck of the bridge. Further,
there is the possibility that the gas in the bubbles is comprised of other gases besides methane.
This could presumably be due the air normally present in local faults and fractures being
displaced by methane intruding under pressure. If this were the case, then the methane
content of the gas in the bubbles would initially contain little or no thermogenic methane, with
relatively sudden increase in methane concentration once intruding methane effectively
purges the fracture
The data available from 3 survey drive-by passes over this area spanned a period of 3.5 hours.
Assuming the measured concentration is consistent from the ground surface to 2 meters
above, the volume of ground level air in the plume area is 4.2 square kilometers X 2m =
4,200,000 square meters x 2m = 8,400,000 cubic meters. A methane concentration increase of
1.8 ppm would require 15.2 cubic meters of methane. Given the 3.5 hours over which this
accumulation occurred, the implied emission rate is 4.3 cubic meters, or 150 cubic feet per
hour. This, however, is a major underestimation of the likely volume of gas being released in
the identified plume. Methane is a low-density gas, about half the density of air.
Consequently, methane will tend to rise in the air relatively rapidly and the lowest methane
concentrations in the vicinity of a surface methane emission will be expected to occur at
ground level. It follows, therefore, that an estimate of the likely methane emission rate in the
identified plume area that includes the vertical extent of the plume would be orders of
-------
magnitude greater than the above estimate (150 cubic feet per hour) based on ground level
methane only. Application of air contaminant diffusion models appropriate to estimating the
full-height methane emission rate was beyond the scope of this effort. The most definitive and
reliable approach would be direct investigation of methane emissions through water and soil
surfaces using the CRDS instrument and appropriate related equipment. However, this
approach requires direct access to the properties on which the methane emissions are
occurring, which could not be obtained for this effort. Further, emissions through soil surfaces
typically are invisible and may occur for prolonged periods with no recognition until
vegetation is damaged or killed by asphyxiation of the roots. Hence, many property owners
may be heavily impacted but be unaware, and, therefore, reluctant to participant in methane
emission survey efforts.
In summary, the methane survey data collected on 8 June 2012 in parts of Leroy, Granville, and
Franklin Townships, Bradford County, Pennsylvania indicated one or more substantial
methane emissions were occurring in an area centered roughly on the intersection of H
Rockwell Road and Route 414. A ground level plume was detected that increased in area
substantially over a period of 3.5 hours, which, when expanded to account for above ground
level methane, suggests large amounts of methane were being emitted to the atmosphere.
Heavily methane contaminated residential water wells occurred in and around the same area,
and documentation indicated heavy contamination had existed for at least several months.
Bubbling gas in Towanda Creek suggested fugitive gas from shale gas wells might be travelling
through faults and fractures, which also carry local ground water, hence, impact local water
wells. Collectively the data and observations suggest natural gas has pervaded an extensive
subsurface area beyond the area where elevated ground-level methane was found during this
survey effort. If that is correct, then more surface emissions should be expected. The issues
and concerns presented in this report require more thorough investigation for confirmation
and quantification.
-------
Appendix H
-------
NEWS
IN FOCUS
Methane leaks erode green
credentials of natural gas
Losses of up to 9% show need for broader data on US gas industry's environmental impact.
BYJEFFTOLLEFSON
Scientists are once again reporting alarm-
ingly high methane emissions from an
oil and gas field, underscoring questions
about the environmental benefits of the boom
in natural-gas production that is transforming
theUS energy system.
The researchers, who hold joint appoint-
ments with the National Oceanic and
Atmospheric Administration (NOAA) and
the University of Colorado in Boulder, first
sparked concern in February 2012 with a
study1 suggesting that up to 4% of the methane
produced at a field near Denver was escaping
into the atmosphere. If methane — a potent
greenhouse gas—is leaking from fields across
the country at similar rates, it could be offset-
ting much of the climate benefit of the ongoing
shift from coal- to gas-fired plants for electric-
ity generation.
Industry officials and some scientists con-
tested the claim, but at an American Geophysi-
cal Union (AGU) meeting in San Francisco,
California, last month, the research team
reported new Colorado data that support the
earlier work, as well as preliminary results from
a field study in the Uinta Basin of Utah sug-
gesting even higher rates of methane leakage
— an eye-popping 9% of the total production.
That figure is nearly double the cumulative lo ss
rates estimated from industry data — which
are already higher in Utah than in Colorado.
"We were expecting to see high methane lev-
els, but I don't think anybody really compre-
hended the true magnitude of what we would
see," says Colm Sweeney, who led the aerial
component of the study as head of the aircraft
programme atNOAA'sEarth SystemResearch
Laboratory inBoulder.
Whether the high leakage rates claimed in
Colorado and Utah are typical across the US
natural-gas industry remains unclear. The
NOAA data represent a "small snapshot" of
a much larger picture that the broader sci-
entific community is now assembling, says
Steven Hamburg, chief scientist at the Envi-
ronmental Defense Fund (EDF) in Boston,
Massachusetts.
The NOAA researchers collected their
data in February as part of a broader analy-
sis of air pollution in the Uinta Basin, using
ground-based equipment and an aircraft to
Natural-gas wells such as this one in Colorado are
increasingly important to the US energy supply.
make detailed measurements of various pol-
lutants, including methane concentrations.
The researchers used atmospheric modelling
to calculate the level of methane emissions
required to reach those concentrations, and
then compared that with industry data on gas
production to obtain the percentage escap-
ing into the atmosphere through venting
and leaks.
The results build on those of the earlier Col-
orado study1 in the Denver-.Tulesburg Basin,
led by NOAA scientist Gabrielle Petron (see
Nature 482,139-140;2012). That study relied
on pollution measurements taken in 2008
on the ground and from a nearby tower, and
estimated a leakage rate that was about twice
as high as official figures suggested. But the
team's methodology for calculating leakage—
based on chemical analysis of the pollutants
— remains in dispute. MichaelLevi, an energy
analyst at the Council on Foreign Relations in
New York, published a peer-reviewed com-
ment2 questioning the findings and presenting
an alternative interpretation of the data that
would align overall leakage rates with previ-
ous estimates.
Petron and her colleagues have a defence of
the Colorado study in press3, and at the AGU
meeting she discussed a new study of the Den-
ver-.Tulesburg Basin conducted with scientists
at Picarro, a gas-analyser manufacturer based
in Santa Clara, California. That study relies
on carbon isotopes to differentiate between
industrial emissions and methane from cows
and feedlots, and the preliminary results line
up with their earlier findings.
A great deal rides on getting the number
right. A study4 published in April by scientists
at the EDF and Princeton University in New
Jersey suggests that shifting to natural gas
from coal-fired generators has immediate cli-
matic benefits as long as the cumulative leak-
age rate from natural-gas production is below
3.2%; the benefits accumulate over time and
are even larger if the gas plants replace older
coal plants. By comparison, the authors note
that the latest estimates from the US Environ-
mental Protection Agency (EPA) suggest that
2.4% of total natural-gas production was lost
toleakagein2009.
To see if that number holds up, the NOAA
scientists are also taking part in a comprehen-
sive assessment of US natural-gas emissions,
conductedby the University of Texas at Austin
and the EDF, with various industry partners.
The initiative will analyse emissions from
the production, gathering, processing, long-
distance transmission and local distribution
of natural gas, and will gather data on the use
of natural gas in the transportation sector. In
addition to scouring through industry data,
the scientists are collecting field measure-
ments at facilities across the country. The
researchers expect to submit the first of these
studies for publication by February, and say
that the others will be complete within ay ear.
In April, the EPA issued standards intended
to reduce air pollution from hydraulic-frac-
turing operations — now standard within the
oil and gas industry — and advocates say that
more can be done, at the state and national lev-
els, to reduce methane emissions. "There are
clearly opportunities to reduce leakage," says
Hamburg. ¦
1. Petron, G. et a/. J Geophys. Res. 117, D04304 (2012).
2. Levi. M. A. J. Geophys. Res. 117, D21203 (2012).
3. Petron, G. etaI J Geophys. Res. (in the press).
4. Alvarez, R. A., Pacala, S. W. Winebrake, J. J.,
Chameides, W. L. & Hamburg, S. P. Proc Natl Acad.
Sci. USA 109, 6435-6440 (2012),
12 I NATURE I VOL 4 9 3 I 3 JANUARY 2013
©2013 Macmiiian Publishers Limited. All rights reserved
-------
Appendix I
-------
NEWS IN FOCUS
SPACEFission-powered FUNDING Japanese BIOMEDICINE Cystic ETHICS The painful
spaceflight gets a boost university puts a donor's fibrosis drag realizes legacy of the Guatemala
atNASAp.141 name in lights B, 143 20-vear-old promise d.145 * • experiments p.148
FUNDING Japanese
university puts a donor's
name in lights p,143
BIOMEDICINE Cystic
fibrosis drag realizes
20-year-old promise p.145
ETHICS The painful
legacy of the Guatemala
experiments p.148
Natural-gas operations in areas such as Wyoming's Jonah Field could release far more methane into the atmosphere than previously thought.
CLIMATECHAN GE
Air sampling reveals high
emissions from gas Held
Methane leaks duringproduction may offset climate benefits ofnatural gas.
BY JEFFTOLLEFSON
When US government scientists
began sampling the air from a
tower north of Denver, Colorado,
they expected urban smog — but not strong
whiffs of what looked like natural gas. They
eventually linked the mysterious pollu-
tion to a nearby natural-gas field, and their
investigation has now produced the first hard
evidence that the cleanest-buming fossil fuel
might not be much better than coal when it
comes to climate change.
Led by researchers at the National Oceanic
and Atmospheric Administration (NOAA)
and the University of C olorado, Boulder, the
study estimates that natural-gas producers in
an area known as the Denver-Julesburg Basin
are losing about 4% of their gas to the atmos-
phere — not including additional losses in
the pipeline and distribution system. This is
more than double the official inventory, but
roughly in line with estimates made in 2011
that have been challenged by industry. And
because methane is some 25 times more effi-
cient than carbon dioxide at trapping heat in
the atmosphere, releases of that magnitude ~
« FE B RUA RY 2 0 12
©2012 Macmillan Publishers Limited. All rights reserved
VO L 4 8 2 I ffAf BRS
-------
IN FOCUS
A LOSING BATTLE
Estimates of methane losses from gas felds near Denver, Colorado, based on air
sampling differ considerably from calculations based on industry activity.
Inventory
of industry
activity
Mobile lab
100 150 200 250
Billion grams of methane per year
300
~ could effectively offset the environmental
edge that natural gas is said to enj oy over other
fossil fuels.
"If we want natural gas to be the cleanest
fossil fuel source, methane emissions have to
be reduced" says Gabrielle Petron, an atmos-
pheric scientist at NOAA and at the University
of Colorado in Boulder, and first author on the
study, currently in press at the Journal of Geo-
physical Research. Emissions will vary depend-
ing on the site, but Petron sees no reason to
think that this particular basin is unique.
"I think we seriously need to look at natural-
gas operations on the national scale."
The results come as a natural-gas boom
hits the United States, driven by a technology
known as hydraulic fracturing, or 'tracking',
that can crack open hard shale formations and
release the natural gas trapped inside. Envi-
ronmentalists are worried about effects such
as water pollution, but the US government is
enthusiastic about tracking. In his State of the
Union address last week, US President Barack
Obama touted natural gas as the key to boost-
ing domestic energy production.
LACK OF DATA
Natural gas emits about half as much
carbon dioxide as coal per unit of energy
when burned, but separate teams at Cornell
University in Ithaca, New York, and at the
US Environmental Protection Agency (EPA)
concluded last year that methane emissions
from shale gas are much larger than pre-
viously thought. The industry and some
academics branded those findings as exag-
gerated, but the debate has been marked by
a scarcity of hard data.
"It's great to get some actual numbers from
the field," says Robert Howarth, a Cornell
researcher whose team raised concerns about
methane emissions from shale-gas drilling in
a pair of papers, one published in April last
year and another last month (R. W. Howarth
etal. Clim. Change Lett. 106, 679-690; 2011;
R. W. Howarth et al. Clim. Change in the
press). "I'm not looking for vindication here,
but [the NOAA] numbers are coming in very
close to ours, maybe a little higher," he says.
Natural gas might still have an advantage
over coal when burned to create electricity,
because gas-fired power plants tend to be newer
and far more efficient than older facilities that
provide the bulk of the country's coal-fired
generation. But only 30% of US gas is used to
produce electricity, Howarth says, with much of
the rest being used for heating, for which there
is no such advantage.
ON THE SCENT
The first clues appeared in 2007, when NOAA
researchers noticed occasional plumes
of pollutants including methane, butane
and propane in air samples taken from a
300-metre-high atmospheric monitoring
tower north of Denver. The NOAA research-
ers worked out the general direction that the
pollution was coming from by monitoring
winds, and in 2008,
the team took advan- «A Ugpartofit
tageolnewequipment justraw
and drove around the ,f . • , ».
, that is leaking
region, sampling the the
air in real time. Iheir „ , „
readings led them to infrastructure.
the Denver-Julesburg
Basin, where more than 20,000 oil and gas
wells have been drilled during the past four
decades.
Most of the wells in the basin are drilled
into 'tight sand' formations that require the
same tracking technology being used in shale
formations. This process involves injecting a
slurry of water, chemicals and sand into wells
at high pressure to fracture the rock and create
veins that can carry trapped gas to the well.
Afterwards, companies need to pump out the
tracking fluids, releasing bubbles of dissolved
gas as well as burps of early gas production.
Companies typically vent these early gases
into the atmosphere for up to a month or more
until the well hits its full
ONATURE.COM stride, at which point it is
Should fracking hooked up to a pipeline,
stop? The team analysed
go.nature.com/adox2r the ratios of various
NATURE I VOL 482 I 9 FEBRUARY
2 0 12
©2012 Macmillan Publishers Limited. All rights reserved
pollutants in the air samples and then tied
that chemical fingerprint back to emissions
from gas-storage tanks built to hold liquid
petroleum gases before shipment. In doing
so, they were able to work out the local emis-
sions that would be necessary to explain the
concentrations that they were seeing in the
atmosphere (see 'A losing battle'). Some of
the emissions come from the storage tanks,
says Petron, "but a big part of it is just raw
gas that is leaking from the infrastructure"
Their range of 2.3 -7.7% lo ss, with a best guess
of 4%, is slightly higher than Cornell's esti-
mate of 2.2-3.8% for shale-gas drilling and
production. It is also higher than calculations
by the EPA, which revised its methodology
last year and roughly doubled the official US
inventory of emissions from the natural-gas
industry over the past decade. Howarth says
the EPA methodology translates to a 2.8% loss.
The Cornell group had estimated that 1.9%
of the gas produced over the lifetime of a typical
shale-gas well escapes through tracking and well
completion alone. NOAAs study doesn't differ-
entiate between gas from fracking and leaks
from any other point in the production process,
but Petron says that fracking clearly contributes
to some of the gas her team measured.
Capturing and storing gases that are being
vented during the fracking process is feasible,
but industry says that these measures are too
costly to adopt. An EPA rule that is due out as
early as April would promote such changes by
regulating emissions from the gas fields.
Officials with America's Natural Gas
Alliance, based in Washington DC, say that
the study is difficult to evaluate based on
a preliminary review, but in a statement to
Nature they add that "the findings raise ques-
tions and warrant a closer examination by the
scientific community" Environmental groups
are pushing the EPA to strengthen pollution
controls in the pending rule, but industry is
pushing to relax many of the requirements.
Many companies are already improving their
practices and reducing emissions throughout
the country, either voluntarily or by regula-
tion, the alliance says.
Not all studies support the higher methane
numbers. Sergey Paltsev, assistant director
for economic research at the Massachusetts
Institute of Technology Energy Initiative in
Cambridge, and his colleagues are gather-
ing information about industry practices for
a study on shale-gas emissions. He says that
their figures are likely to come in well below
even the lower EPA estimate. He calls the
NOAA results "surprising" and questions how
representative the site is.
Petron says that more studies are needed
using industry inventories and measurements
of atmospheric concentrations. "We will never
get the same numbers," she says, "but if we can
get close enough that our ranges overlap in a
meaningful way, then we can say we under-
stand the process." ¦
-------
Appendix J
-------
Anthropogenic emissions of methane in the
United States
Scot M. Miller3,1, Steven C. Wofsy3, Anna M. Michalakb, Eric A. Kortc, Arlyn E. Andrewsd, Sebastien C. Biraud®,
Edward J. Dlugokenckyd, Janusz Eluszkiewicz', Marc L. Fischer9, Greet Janssens-Maenhouth, Ben R. Miller',
John B. Miller', Stephen A. Montzkad, Thomas Nehrkornf, and Colm Sweeney'
"Department of Earth and Planetary Sciences, Harvard University, Cambridge, MA 02138; bDepartment of Global Ecology, Carnegie Institution for Science,
Stanford, CA 94305; 'Department of Atmospheric, Ocean, and Space Sciences, University of Michigan, Ann Arbor, Ml 48109; dGlobal Monitoring Division,
Earth System Research Laboratory, National Oceanic and Atmospheric Administration, Boulder, CO 80305; 'Earth Sciences Division, and Environmental
Energy Technologies Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720; 'Atmospheric and Environmental Research, Lexington, MA 02421;
institute for Environment and Sustainability, European Commission Joint Research Centre, 21027 Ispra, Italy; and Cooperative Institute for Research in
Environmental Sciences, University of Colorado Boulder, Boulder, CO 80309
Edited by Mark H. Thiemens, University of California, San Diego, La Jolla, CA, and approved October 18, 2013 (received for review August 5, 2013)
This study quantitatively estimates the spatial distribution of
anthropogenic methane sources in the United States by combining
comprehensive atmospheric methane observations, extensive
spatial datasets, and a high-resolution atmospheric transport
model. Results show that current inventories from the US Envi-
ronmental Protection Agency (EPA) and the Emissions Database
for Global Atmospheric Research underestimate methane emis-
sions nationally by a factor of —1.5 and ~1.7, respectively. Our
study indicates that emissions due to ruminants and manure are
up to twice the magnitude of existing inventories. In addition, the
discrepancy in methane source estimates is particularly pro-
nounced in the south-central United States, where we find total
emissions are —2.7 times greater than in most inventories and
account for 24 ± 3% of national emissions. The spatial patterns
of our emission fluxes and observed methane-propane correla-
tions indicate that fossil fuel extraction and refining are major
contributors (45 ± 13%) in the south-central United States. This
result suggests that regional methane emissions due to fossil fuel
extraction and processing could be 4.9 ± 2.6 times larger than in
EDGAR, the most comprehensive global methane inventory. These
results castdoubtontheUSEPA'srecentdecisiontodownscaleits
estimate of national natural gas emissions by 25-30%. Overall, we
conclude that methane emissions associated with both the animal
husbandry and fossil fuel industries have larger greenhouse gas
impacts than indicated by existing inventories.
climate change policy | geostatistical inverse modeling
Methane (CH4) is the second most important anthropogenic
greenhouse gas, with approximately one third the total
radiative forcing of carbon dioxide (1). CH4 also enhances the
formation of surface ozone in populated areas, and thus
higher global concentrations of CH4 may significantly in-
crease ground-level ozone in the Northern Hemisphere (2).
Furthermore, methane affects the ability of the atmosphere to
oxidize other pollutants and plays a role in water formation
within the stratosphere (3).
Atmospheric concentrations of CH4 [ ~ 1,800 parts per billion
(ppbV| are currently much higher than preindustrial levels
(~ 680-715 ppb) (1 ,'4). The global atmospheric burden started to
rise rapidly in the 18th century and paused in the 1990s. Methane
levels began to increase again more recently, potentially from
a combination of increased anthropogenic and/or tropical wet-
land emissions (5—7). Debate continues, however, over the cau-
ses behind these recent trends (7, 8).
Anthropogenic emissions account for 50-65% of the global
CH4 budget of — 395—427 teragrams of carbon per year (TgCy)
(526-569 Tg CH4) (7, 9). The US Environmental Protection
Agency (EPA) estimates the principal anthropogenic sources in
the United States to be (in order of importance) (i) livestock
(enteric fermentation and manure management), (ii) natural gas
www.pnas.org/cgi/do i/10.1073/pnas. 1314392110
production and distribution, (iii) landfills, and (iv) coal mining
(10). EPA assesses human-associated emissions in the United
States in 2008 at 22.1 TgC, roughly 5% of global emissions (10).
The amount of anthropogenic CH4 emissions in the US and
attributions by sector and region are controversial (Fig. 1).
Bottom-up inventories from US EPA and the Emissions Data-
base for Global Atmospheric Research (EDGAR) give totals
ranging from 19.6 to 30 TgCy-1 (10, 11). The most recent EPA
and EDGAR inventories report lower US anthropogenic emis-
sions compared with previous versions (decreased by 10% and
35%, respectively) (10, 12); this change primarily reflects lower,
revised emissions estimates from natural gas and coal production
Fig. SI. However, recent analysis of CH4 da{a from aircraft esti-
mates a higher budget of 32.4 ± 4.5 TgCy for 2004 (13). Fur-
thermore, atmospheric observations indicate higher emissions in
natural gas production areas (14—16); a steady 20-y increase in the
number of US wells and newly-adopted horizontal drilling techni-
ques may have further increased emissions in these regions (17,18).
These disparities among bottom-up and top-down studies
suggest much greater uncertainty in emissions than typically
reported. For example, EPA cites an uncertainty of only ±13%
for the for United States (10). Independent assessments of bot-
tom-up inventories give error ranges of 50—100% (19, 20), and
Significance
Successful regulation of greenhouse gas emissions requires
knowledge of current methane emission sources. Existing state
regulations in California and Massachusetts require —15%
greenhouse gas emissions reductions from current levels by
2020. However, government estimates for total US methane
emissions may be biased by 50%, and estimates of individual
source sectors are even more uncertain. This study uses at-
mospheric methane observations to reduce this level of un-
certainty. We find greenhouse gas emissions from agriculture
and fossil fuel extraction and processing (i.e., oil and/or natural
gas) are likely a factor of two or greater than cited in existing
studies. Effective national and state greenhouse gas reduction
strategies may be difficult to develop without appropriate
estimates of methane emissions from these source sectors.
Author contributions: S.C.W., and A.M.M. designed research; A.E.A., S.C.B.,
E.J.D., J.E., M.L.F., G.J.-M., B.R.M., SAM., T.N., and C.S. performed research; S.M.M.
analyzed data; S.M.M., S.C.W., A.M.M., and E.A.K. wrote the paper; A.E.A., S.C.B., E.J.D.,
M.L.F., B.R.M., J.B.M., S.A.M., and C.S. collected atmospheric methane data; and J.E. and T.N.
developed meteorological simulations using the Weather Research and Forecasting model.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1To whom correspondence should be addressed. E-mail: scot.m.miller@gmail.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi: 10.
1073/pnas. 1314392110/-/DCSu pp I e mental.
PNAS Early Edition | 1 of 5
-------
US budget
m
CO
o
CO
m
CM
O
O)
32.4
30.0
22.1
19.6
TX/OK/KS
budget
33.4
3.8
3.0 3.0 =
¦¦~
XC&X
0o°0e/<
-------
a covariance function to describe the spatial and temporal cor-
relation of the stochastic component and optimizes its spatial
and temporal distribution simultaneously with the optimization
of the activity datasets in the deterministic component (SI Text,
Fig. S5) (26—28). Because of the stochastic component, the final
emissions estimate can have a different spatial and temporal
distribution from any combination of the activity data.
If the observation network is sensitive to a broad array of
different source sectors and/or if the spatial activity maps are
effective at explaining those sources, many activity datasets will
be included in the deterministic model. If the deterministic
model explains the observations well, the magnitude of CH4
emissions in the stochastic component will be small, the assign-
ment to specific sectors will be unambiguous, and uncertainties
in the emissions estimates will be small. This result is not the case
here, as discussed below (see Results).
A number of previous studies used top-down methods to
constrain anthropogenic CH4 sources from global (29—33) to
regional (13—15, 34—38) scales over North America. Most regional
studies adopted one of three approaches: use a simple box model
to estimate an overall CH4 budget (14), estimate a budget using
the relative ratios of different gases (15, 37—3.9), or estimate
scaling factors for inventories by region or source type: (13, 34—
36). The first two methods do not usually give explicit in-
formation about geographic distribution. The last approach
provides information about the geographic distribution of sour-
ces, but results hinge on the spatial accuracy of the underlying
regional or sectoral emissions inventories (40).
Here, we are able to provide more insight into the spatial
distribution of emissions; like the scaling factor method above,
we leverage spatial information about source sectors from an
existing inventory, but in addition we estimate the distribution of
emissions where the inventory is deficient. We further bolster
attribution of regional emissions from the energy industry using
the observed correlation of CH4 and propane, a gas not pro-
duced by biogenic processes like livestock and landfills.
Results
Spatial Distribution of CH4 Emissions. Fig. 3 displays the result of
the 2-y mean of the monthly CH4 inversions and differences from
the EDGAR 4.2 inventory. We find emissions for the United
States that are a factor of 1.7 larger than the EDGAR inventory.
The optimized emissions estimated by this study bring the model
closer in line with the observations (Fig. 4, Figs. S6 and S7).
Posterior emissions fit the CH4 observations [R2 = 0:64, root
mean square error (RMSE) = 31 ppb] much better than EDGAR
v4.2 (R2=0:23, RMSE = 49 ppb). Evidently, the spatial distri-
bution of EDGAR sources is inconsistent with emissions patterns
implied by the CH4 measurements and associated footprints.
Several diagnostic measures preclude the possibility of major
systematic errors in WRF—STILT. First, excellent agreement
between the model and measured vertical profiles from aircraft
implies little bias in modeled vertical air mixing (e.g., boundary-
layer heights) (Fig. 4). Second, the monthly posterior emissions
estimated by the inversion lack statistically significant seasonality
(Fig. S8). This result implies that seasonally varying weather
patterns do not produce detectable biases in WRF—STILT. SI
Text discusses possible model errors and biases in greater detail.
GH4 observations are sparse over parts of the southern and
central East Coast and in the Pacific Northwest. Emissions
estimates for these regions therefore rely more strongly on the
deterministic component of the flux model, with weights
constrained primarily by observations elsewhere. Therefore,
emissions in these areas, including from coal mining, are
poorly constrained (SI Text).
Contribution of Different Source Sectors. Only two spatial activity
datasets from EDGAR 4.2 are selected through the BIC as
meaningful predictors of CH4 observations over the United
States: population densities of humans and of ruminants (Table
SI), Some sectors are eliminated by the BIG because emissions
are situated far from observation sites (e.g., coal mining in West
Virginia or Pennsylvania), making available CH4 data insensitive
to these predictors. Other sectors may strongly affect observed
concentrations but are not selected, indicating that the spatial
datasets from EDGAR are poor predictors for the distribution of
observed concentrations (e.g., oil and natural gas extraction and
oil refining). Sources from these sectors appear in the stochastic
component of the GIM (SI Text).
The results imply that existing inventories underestimate emis-
sions from two key sectors: ruminants and fossil fuel extraction
and/or processing, discussed in the remainder of this section.
We use the optimized ruminant activity dataset to estimate the
magnitude of emissions with spatial patterns similar to animal
husbandry and manure. Our corresponding US budget of 12.7 ±
5.0 TgC'\ is nearly twice that of EDGAR and EPA (6.7 and
7.0, respectively). The total posterior emissions estimate over the
northern plains, a region with high ruminant density but little
fossil fuel extraction, further supports the ruminant estimate
(Nebraska, Iowa, Wisconsin, Minnesota, and South Dakota).
Our total budget for this region of 3.4 ± 0.7 compares with 1.5
TgC-y in EDGAR. Ruminants and agriculture may also be
>.04
0.02
0.00
I - -0.02
U -0.04
Hmol rrr2 s~1
Fig. 3. The 2-y averaged CH4 emissions estimated in this study (A) compared against the commonly used EDGAR 4.2 inventory (B and C). Emissions estimated
in this study are greater than in EDGAR 4.2, especially near Texas and California.
-130
-60 -130
-60-130
A This study (2007-2008 average)
B
EDGARv4.2 inventory
C This study minus EDGARv4.2
Miller et al.
PNAS Early Edition | 3 of 5
-------
All sites
Measurements
Boundary
Wetland model
Edgar v4.2
Posterior
emissions
—I 1 1 1—
1820 1840 1860 1880
Ponca City, Oklahoma
(SGP)
Cape May, NJ
(CMA)
1820
1860
1900
1 1 1 1 T
1820 1840 1860 1880 1900
1820 1840 1860
1880
CH (ppb)
Fig. 4. A model-measurement comparison at several regular NOAA/DOE aircraft monitoring sites (averaged over 2007-2008). Plots include the measure-
ments; the modeled boundary condition; the summed boundary condition and wetland contribution (from the Kaplan model); and the summed boundary,
wetland, and anthropogenic contributions (from EDGAR v4.2 and the posterior emissions estimate).
partially responsible for high emissions over California (41).
EDGAR activity datasets are poor over California (42), but
several recent studies (34, 36—38, 41) have provided detailed top-
down emissions estimates tor the state using datasets from state
agencies.
Existing inventories also greatly underestimate CH4 sources
from the south-central United States (Fig. 3). We find the total
CH4 source from Texas, Oklahoma, and Kansas to be 8.1 ±0.96
TgC -y_1, a factor of 2.7 higher than the EDGAR inventory. These
three states alone constitute ~24 ± 3% of the total US anthro-
pogenic CH4 budget or 3.7% of net US greenhouse gas emissions
[in CO2 equivalents (10)].
Texas and Oklahoma were among the top five natural gas pro-
ducing states in the country in 2007 (18), and aircraft observations of
alkanes indicate that the natural gas and/or oil industries play a sig-
nificant role in regional CH4 emissions. Concentrations of propane
(CjHs), a tracer of fossil hydrocarbons (43), are strongly correlated
with CH4 at NOAA/DOE aircraft monitoring locations over Texas
and Oklahoma (R2 = 0:72) (Fig. 5). Correlations are much weaker at
other locations in North America (R2 =0:11 to 0.64).
We can obtain an approximate CH4 budget for fossil-fuel ex-
traction in the region by subtracting the optimized contributions
associated with ruminants and population from the total emis-
sions. The residual (Fig. S4C) represents sources that have
spatial patterns not correlated with either human or ruminant
density in EDGAR. Our budget sums to 3.7 ± 2.0 TgC-y-1,
a factor of 4.9 ± 2.6 larger than oil and gas emissions in ED-
GAR v4.2 (0.75 TgCy ) and a factor of 6.7 ± 3.6 greater than
EDGAR sources from solid waste facilities (0.55 TgCy :), the
two major sources that may not be accounted for in the de-
terministic component. The population component likely cap-
tures a portion of the solid waste sources so this residual methane
budget more likely represents natural gas and oil emissions than
landfills. SI Text discusses in detail the uncertainties in this sector-
based emissions estimate. We currently do not have the detailed,
accurate, and spatially resolved activity data (fossil fuel extraction
and processing, ruminants, solid waste) that would provide more
accurate sectorial attribution.
Katzenstein et al. (2003) (14) were the first to report large
regional emissions of CH4 from Texas, Oklahoma, and Kansas;
they cover an earlier time period (1999—2002) than this study.
They used a box model and 261 near-ground CH4 measurements
taken over 6 d to estimate a total Texas-Oklahoma—Kansas CH4
budget (from all sectors) of 3.8 ± 0.75 TgC y-1. We revise their
1800
i r
1900 2000
"i i 1 r
2100 2200 1800 1900 2000
Methane (CH ppb)
1 r
2100 2200
Fig. 5. Correlations between propane and ChU at NOAA/DOE aircraft observation sites in Oklahoma (A) and Texas (ES) over 2007-2012. Correlations are higher in
these locations than at any other North American sites, indicating large contributions of fossil fuel extraction and processing to ChU emitted in this region.
4 of 5 | www.pnas.org/cgi/doi/10.1073/pnas.1314392110
Miller et al.
-------
estimate upward by a factor of two based on the inverse model
and many more measurements from different platforms over two
full years of data. SI Text further compares the CH4 estimate in
Katzenstein et al. and in this study.
Discussion and Summary
This study combines comprehensive atmospheric data, diverse
datasets from the EDGAR inventory, and an inverse modeling
framework to derive spatially resolved CH4 emissions and
information on key source sectors. We estimate a mean annual
US anthropogenic CH4 budget for 2007 and 2008 of 33.4 ± 1.4
TgC y-1 or ~7-8% of the total global CH4 source. This estimate
is a factor of 1.5 and 1.7 larger than EPA and EDGAR v4.2,
respectively. CH4 emissions from Texas, Oklahoma, and Kansas
alone account for 24% of US methane emissions, or 3.7% of the
total US greenhouse gas budget.
The results indicate that drilling, processing, and refining activi-
ties over the south-central United States have emissions as much as
4.9 ± 2.6 times larger than EDGAR, and livestock operations across
the US have emissions approximately twice that of recent in-
ventories. The US EPA recently decreased its CH4 emission factors
for fossil fuel extraction and processing by 25—30%) (for 1990—2011)
(10), but we find that CH4 data from across North America instead
indicate the need for a larger adjustment of the opposite sign.
1. Butler J (2012) The NOAA annual greenhouse gas index (AGGI). Available at http://
www.esrl.noaa.gov/gmd/aggi/. Accessed November 4, 2013.
2. Fiore AM, et al. (2002) Linking ozone pollution and climate change: The case for
controlling methane. Geophys Res Lett 29:1919.
3. Jacob D (1999) Introduction to Atmospheric Chemistry (Princeton Univ Press, Prince-
ton).
4. Mitchell LE, Brook EJ, Sowers T, McConnell JR, Taylor K (2011) Multidecadal variability
of atmospheric methane, 1000-1800 CE. J Geophys Res Biogeosci 116:G02007.
5. Dlugokencky EJ, et al. (2009) Observational constraints on recent increases in the
atmospheric ChU burden. Geophys Res Lett 36:L18803.
6. Sussmann R, Forster F, Rettinger M, Bousquet P (2012) Renewed methane increase for
five years (2007-2011) observed by solar FTIR spectrometry. Atmos Chem Phys 12:
4885-4891.
7. Kirschke S, et al. (2013) Three decades of global methane sources and sinks. Nat
Geosci 6:813-823.
8. Wang JS, et al. (2004) A 3-D model analysis of the slowdown and interannual vari-
ability in the methane growth rate from 1988 to 1997. Global Biogeochem Cycles 18:
GB3011.
9. Ciais P, et al. (2013) Carbon and Other Biogeochemical Cycles: Final Draft Underlying
Scientific Technical Assessment (I PCC Secretariat, Geneva).
10. US Environmental Protection Agency (2013) Inventory of U.S. Greenhouse Gas Emis-
sions and Sinks: 1990-2011, Technical Report EPA 430-R-13-001 (Environmental Pro-
tection Agency, Washington).
11. Olivier JGJ, Peters J (2005) CO2 from non-energy use of fuels: A global, regional and na-
tional perspective based on the IPCC Tier 1 approach. Resour Conserv Recycling 45:210-225.
12. European Commission Joint Research Centre, Netherlands Environmental Assessment
Agency (2010) Emission Database for Global Atmospheric Research (EDGAR), Release
Version 4.2. Available at http://edgar.jrc.ec.europa.eu. Accessed November 4, 2013.
13. Kort EA, et al. (2008) Emissions of ChUand N2O over the United States and Canada
based on a receptor-oriented modeling framework and COBRA-NA atmospheric ob-
servations. Geophys Res Lett 35: L18808.
14. Katzenstein AS, Doezema LA, Simpson I J, Blake DR, Rowland FS (2003) Extensive re-
gional atmospheric hydrocarbon pollution in the southwestern United States.
Proc Natl Acad Sci USA 100(21): 11975-11979.
15. Petron G, et al. (2012) Hydrocarbon emissions characterization in the Colorado Front
Range: A pilot study. J Geophys Res Atmos 117:D04304.
16. Karion A, et al. (2013) Methane emissions estimate from airborne measurements over
a western United States natural gas field. Geophys Res Lett 40:4393-4397.
17. Howarth RW, Santoro R, Ingraffea A (2011) Methane and the greenhouse-gas foot-
print of natural gas from shale formations. Clim Change 106:679-690.
18. US Energy Information Administration (2013) Natural Gas Annual 2011, Technical
report (US Department of Energy, Washington).
19. National Research Council (2010) Verifying Greenhouse Gas Emissions: Methods to
Support International Climate Agreements (National Academies Press, Washington).
20. Dlugokencky EJ, Nisbet EG, Fisher R, Lowry D (2011) Global atmospheric meth-
ane: Budget, changes and dangers. Philos Trans A Math Phys Eng Sci 369(1943):
2058-2072.
21. Lin JC, et al. (2003) A near-field tool for simulating the upstream influence of at-
mospheric observations: The Stochastic Time-Inverted Lagrangian Transport (STILT)
model. J Geophys Res Atmos 108(D16):4493.
ACKNOWLEDGMENTS. For advice and support, we thank Roisin Commane,
Elaine Gottlieb, and Matthew Hayek (Harvard University); Robert Harriss
(Environmental Defense Fund); Hanqin Tian and Bowen Zhang (Auburn Uni-
versity); Jed Kaplan (Ecole Polytechnique Federale de Lausanne); Kimberly
Mueller and Christopher Weber (Institute for Defense Analyses Science and
Technology Policy Institute); Nadia Oussayef; and Gregory Berger. In addi-
tion, we thank the National Aeronautics and Space Administration (NASA)
Advanced Supercomputing Division for computing help; P. Lang, K. Sours,
and C. Siso for analysis of National Oceanic and Atmospheric Administration
(NOAA) flasks; and B. Hall for calibration standards work. This work was
supported by the American Meteorological Society Graduate Student Fel-
lowship/Department of Energy (DOE) Atmospheric Radiation Measurement
Program, a DOE Computational Science Graduate Fellowship, and the
National Science Foundation Graduate Research Fellowship Program.
NOAA measurements were funded in part by the Atmospheric Composi-
tion and Climate Program and the Carbon Cycle Program of NOAA's
Climate Program Office. Support for this research was provided by NASA
Grants NNX08AR47G and NNX11AG47G, NOAA Grants NA090AR4310122
and NA11OAR4310158, National Science Foundaton (NSF) Grant ATM-
0628575, and Environmental Defense Fund Grant 0146-10100 (to Harvard
University). Measurements at Walnut Grove were supported in part by
a California Energy Commission Public Interest Environmental Research
Program grant to Lawrence Berkeley National Laboratory through the US
Department of Energy under Contract DE-AC02-05CH11231. DOE flights
were supported by the Office of Biological and Environmental Research
of the US Department of Energy under Contract DE-AC02-05CH11231 as
part of the Atmospheric Radiation Measurement Program (ARM), ARM
Aerial Facility, and Terrestrial Ecosystem Science Program. Weather Re-
search and Forecasting-Stochastic Time-Inverted Lagrangian Transport
model development at Atmospheric and Environmental Research has
been funded by NSF Grant ATM-0836153, NASA, NOAA, and the US
intelligence community.
22. Nehrkorn T, et al. (2010) Coupled Weather Research and Forecasting-Stochastic Time-
Inverted Lagrangian Transport (WRF-STILT) model. Meteorol Atmos Phys 107:51-64.
23. NOAA ESRL (2013) Carbon Cycle Greenhouse Gas Group Aircraft Program. Available
at http://www.esrl.noaa.gov/gmd/ccgg/aircraft/index.html. Accessed November 4, 2013.
24. Biraud SC, et al. (2013) A multi-year record of airborne CO2 observations in the US
southern great plains. Atmos Meas Tech 6:751-763.
25. Pan LL, et al. (2010) The Stratosphere-Troposphere Analyses of Regional Transport
2008 Experiment. Bull Am Meteorol Soc 91:327-342.
26. Kitanidis PK, Vomvoris EG (1983) A geostatistical approach to the inverse problem
in groundwater modeling (steady state) and one-dimensional simulations. Water
ResourRes 19:677-690.
27. Michalak A, Bruhwiler L, Tans P (2004) A geostatistical approach to surface flux es-
timation of atmospheric trace gases. J Geophys Res Atmos 109(D14):D14109.
28. Gourdji SM, etal. (2012) North American C02 exchange: Inter-comparison of modeled
estimates with results from a fine-scale atmospheric inversion. Biogeosciences 9:
457-475.
29. Chen YH, Prinn RG (2006) Estimation of atmospheric methane emissions between
1996 and 2001 using a three-dimensional global chemical transport model. J Geophys
Res Atmos 111 (D10): D10307.
30. Meirink JF, et al. (2008) Four-dimensional variational data assimilation for inverse
modeling of atmospheric methane emissions: Analysis of SCIAMACHY observations.
J Geophys Res Atmos 113(D17):D17301.
31. Bergamaschi P, et al. (2009) Inverse modeling of global and regional CH4 emissions
using SCIAMACHY satellite retrievals. J Geophys Res Atmos 114(D22):D22301.
32. Bousquet P, et al. (2011) Source attribution of the changes in atmospheric methane
for 2006-2008. Atmos Chem Phys 11:3689-3700.
33. Monteil G, et al. (2011) Interpreting methane variations in the past two decades using
measurements of CH4 mixing ratio and isotopic composition. Atmos Chem Phys 11:
9141-9153.
34. Zhao C, et al. (2009) Atmospheric inverse estimates of methane emissions from central
California. J Geophys Res Atmos 114(D16):D16302.
35. Kort EA, et al. (2010) Atmospheric constraints on 2004 emissions of methane and
nitrous oxide in North America from atmospheric measurements and receptor-ori-
ented modeling framework. J Integr Environ Sci 7:125-133.
36. Jeong S, et al. (2012) Seasonal variation of CH4 emissions from central California.
J Geophys Res 117:D11306.
37. Peischl J, et al. (2012) Airborne observations of methane emissions from rice culti-
vation in the Sacramento Valley of California. J Geophys Res Atmos 117(D24):D00V25.
38. Wennberg PO, et al. (2012) On the sources of methane to the Los Angeles atmo-
sphere. Environ Sci Technol 46( 17):9282—9289.
39. Miller JB, et al. (2012) Linking emissions of fossil fuel C02 and other anthropogenic
trace gases using atmospheric 14C02. J Geophys Res Atmos 117(D8):D08302.
40. Law RM, Rayner PJ, Steele LP, Enting IG (2002) Using high temporal frequency data
for C02 inversions. Global Biogeochem Cycles 16(4):1053.
41. Jeong S, et al. (2013) A multitower measurement network estimate of California's
methane emissions. J Geophys Res Atmos, 10.1002/jgrd.50854.
42. Xiang B, et al. (2013) Nitrous oxide (N20) emissions from California based on 2010
CalNex airborne measurements. J Geophys Res Atmos 118(7):2809-2820.
43. Koppmann R (2008) Volatile Organic Compounds in the Atmosphere (Wiley,
Singapore).
Miller et al.
PNAS Early Edition | 5 of 5
------- |